University of Liège

Applied Sciences Faculty
Centre Spatial de Liège
Thermal Design of the Oufti-1
nanosatellite
Lionel Jacques
Thesis submitted in partial fulfillment of the requirements for the degree of
Master in Aerospace Engineering
Advisor : Prof. Pierre Rochus
Academic Year 2008 2009
1
Space isn’t remote at all. It’s only an
hour’s drive away, if your car could go
straight upwards.
Sir Fred Hoyle
ii
Acknowledgments
I
wish to thank all the people who helped me, during the whole year, to
carry out this master thesis.
I express my sincere thanks to Prof. Pierre Rochus for his precious
advice and encouragement and for giving me the opportunity to enter the
very interesting world of spacecraft thermal control.
I am greatly indebted to Tanguy Thibert for his guidance, suggestions,
useful feedbacks and all the support he has given throughout my work.
I expressed also my sincere gratitude to all the people working at the
Centre Spatial de Liège for their availability and support in helping me
conducting the different tests.
I also wish to thank Philippe Poinas for his valuable advice and help
in developing OUFTI-1’s thermal model.
My special thanks goes to all the OUFTI-1 student and management
team, especially to Amandine Denis and Jonathan Pisane who managed
the project with an incomparable team-spirit. Particularly, I am thankful to
Prof. Gaetan Kerschen for his valuable suggestions and recommendations
and to Vincent Beukelaers who helped me during the tests.
Last but not least I thank my family and I thank you, Géraldine, for
encouraging, supporting me and not being weary of always hearing me
talking about OUFTI-1. . .
Again, many thanks to all of you, for keeping our eyes turned toward
the sky. . .
Harzé, June 15, 2009.
iii
Abstract
OUFTI-1 is the first nanosatellite developed at the University of Liège
and even the first one ever made in Belgium. This student project takes
place within the framework of a long-term goal program called LEODIUM
(Liège in Latin). The goal of this program is to provide hands-on expe-
rience to engineering students through the development of a series of
nanosatellite for scientific experiments in cooperation with space indus-
tries of the region of Liège.
OUFTI-1 will be the first satellite ever equipped with one of the lat-
est digital amateur radio communication protocol : the D-STAR protocol.
This technology represents one of the three payload of OUFTI-1, since one
target of the mission is to test this new protocol in space. The two other
payloads are high efficiency solar cells provided by Azur Space and an
innovative electrical power system developed with Thales Alenia Space
ETCA.
OUFTI-1 will hopefully be launched on the new European launcher
Vega with eight other student nanosatellites.
This present thesis focuses on the thermal design of OUFTI-1 whose
goal is to guarantee all components are functioning within their allow-
able temperature range. With this in mind, different thermal model of
increasing complexity are developed within both Matlab/Simulink and
ESATAN/ESARAD environments. based on their results, proper measures
will be taken to ensure all the components works indeed in their allowable
range of temperatures.
Keywords : [ OUFTI-1, CubeSat, satellite, Thermal, LEODIUM, ESA-
TAN, ESARAD]
iv
Contents
Contents v
List of Figures viii
List of Tables xii
Thesis outline 1
1 OUFTI-1 project, mission and nanosatellite 2
1.1 The Project Story . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 The Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 The Satellite . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 The Subsystems . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 OUFTI-1’s anatomy . . . . . . . . . . . . . . . . . . . . . . . 11
2 The Thermal Control Subsystem 13
2.1 Space heat transfer . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.1 Conduction . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.2 Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 The thermal environment . . . . . . . . . . . . . . . . . . 16
2.3 Thus, why Thermal Control ? and how ? . . . . . . . . . 17
2.4 Thermal control systems of other CubeSats . . . . . . . 18
3 Preliminary thermal analysis 20
3.1 Notations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Thermal Requirements . . . . . . . . . . . . . . . . . . . . . 21
3.3 Thermal environment related to OUFTI-1’s orbit . . . . 22
3.3.1 The orbit . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.3.2 Solar flux . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3.3 Albedo flux . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3.4 Earth infrared flux . . . . . . . . . . . . . . . . . . . . . 26
3.3.5 Aerothermal flux . . . . . . . . . . . . . . . . . . . . . . 27
3.3.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 Cyclic Transient Model . . . . . . . . . . . . . . . . . . . . 29
3.4.1 Implementation . . . . . . . . . . . . . . . . . . . . . . . 29
3.4.2 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.5.1 Hot Case . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.5.2 Cold Case . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.6 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . 34
3.7 Advanced Simulink Model . . . . . . . . . . . . . . . . . . 35
4 Simplified Thermal Model 38
4.1 Thermal modeling with ESATAN & ESARAD . . . . . . . 38
4.2 Geometric Mathematical Model . . . . . . . . . . . . . . 40
4.2.1 Nodal breakdown . . . . . . . . . . . . . . . . . . . . . 40
4.2.2 Thermo-optical properties . . . . . . . . . . . . . . . . . 40
4.2.3 Orbit & attitude . . . . . . . . . . . . . . . . . . . . . . 41
4.2.4 Radiative coupling . . . . . . . . . . . . . . . . . . . . . 42
4.3 Thermal Mathematical Model . . . . . . . . . . . . . . . . 44
4.3.1 The structure . . . . . . . . . . . . . . . . . . . . . . . . 44
4.3.2 The solar cells . . . . . . . . . . . . . . . . . . . . . . . 46
4.3.3 The PCBs stack . . . . . . . . . . . . . . . . . . . . . . . 47
4.3.4 The batteries . . . . . . . . . . . . . . . . . . . . . . . . 48
4.4 Worst cases definition . . . . . . . . . . . . . . . . . . . . . 49
4.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.5.1 Hot case . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.5.2 Cold case . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.5.3 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . 54
5 Measurements 56
5.1 Why and which measurements ? . . . . . . . . . . . . . . . 56
5.2 The Battery . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.2.1 Experimental setup . . . . . . . . . . . . . . . . . . . . . 56
5.2.2 The model . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.2.3 Model adjustment & results . . . . . . . . . . . . . . . . 58
5.3 aluminum frame emissivity . . . . . . . . . . . . . . . . . . 59
5.3.1 Anodizing vs Alodine . . . . . . . . . . . . . . . . . . . 59
5.3.2 Measuring emissivity with a thermographic camera . . . 60
5.3.3 Experimental setup . . . . . . . . . . . . . . . . . . . . . 62
5.3.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.4 aluminum frame contact resistance . . . . . . . . . . . . 64
6 Detailed Thermal Model 67
6.1 Geometric Mathematical Model . . . . . . . . . . . . . . 67
6.1.1 Nodal breakdown . . . . . . . . . . . . . . . . . . . . . 67
6.1.2 Thermo-optical properties . . . . . . . . . . . . . . . . . 70
6.1.3 ESARAD considerations . . . . . . . . . . . . . . . . . . 70
6.2 Thermal Mathematical Model . . . . . . . . . . . . . . . . 70
6.2.1 Conductive network . . . . . . . . . . . . . . . . . . . . 70
6.2.2 Internal power . . . . . . . . . . . . . . . . . . . . . . . 74
6.3 Checking the model . . . . . . . . . . . . . . . . . . . . . . 77
6.3.1 GMM . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.3.2 TMM . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.4.1 Absorbed powers . . . . . . . . . . . . . . . . . . . . . . 80
6.4.2 Cold Case . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.4.3 Hot Case 1 . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.4.4 Hot Case 2 . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.5 Parametric analysis and design . . . . . . . . . . . . . . . 85
6.5.1 Cold case issues . . . . . . . . . . . . . . . . . . . . . . 86
6.5.2 Hot case issues . . . . . . . . . . . . . . . . . . . . . . . 94
7 Testing 101
7.1 Test and model philosophy . . . . . . . . . . . . . . . . . . 101
7.2 Tests specifications . . . . . . . . . . . . . . . . . . . . . . . 102
7.3 Test set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Conclusions 113
Appendix 115
A Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
B Aluminium frame faces links . . . . . . . . . . . . . . . . . 117
C Analytical computation of VF between parallel rect-
angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
D Advanced Simulink Thermal Model . . . . . . . . . . . . 121
E 2D conduction lumped parameter method . . . . . . . . 124
Bibliography 126
List of Figures
1.1 Mission description : Doppler-Effect compensated zones . . 5
1.2 OUFTI-1’s three payloads . . . . . . . . . . . . . . . . . . . . 6
1.3 On the left, the CubeSat Kit Structure, on the right, the P-POD 7
1.4 Earth’s magnetic field orientation on the orbit[25] . . . . . . 7
1.5 OUFTI-1’s test RX PCB[36],[28] . . . . . . . . . . . . . . . . . 8
1.6 Main EPS engineering model . . . . . . . . . . . . . . . . . . 8
1.7 OUFTI-1’s antenna deployment mechanism panel[63] . . . . 9
1.8 Snapshot of the Simulator developed by V. Beukelaers in
Simulink[9] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.9 OUFTI-1’s second OBC PCB[19] . . . . . . . . . . . . . . . . 10
1.10 Finite element modal analysis with SamcefField[42] . . . . . 10
1.11 OUFTI-1 exploded view . . . . . . . . . . . . . . . . . . . . . 12
2.1 Planck’s black body radiation curves for different tempera-
tures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Thermal environment of a satellite in LEO . . . . . . . . . . 17
3.1 Classical orbital elements . . . . . . . . . . . . . . . . . . . . 22
3.2 Simulink two-body propagator for incident flux computa-
tion : main window (top) and Albedo and IR calculation
box (bottom) . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Incident albedo irradiation on a surface element in Earth
orbit [24] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.4 Fitting of the Gilmore abacus curves with the formula 3.2 . 26
3.5 Nominal Earth IR and Albedo fluxes seen by a surface
pointing to the Earth on OUFTI-1’s orbit . . . . . . . . . . . 27
3.6 Infrared fluxes incident on each face and effective area ratio 28
3.7 Effective area ratio . . . . . . . . . . . . . . . . . . . . . . . . 28
3.8 Cyclic transient Matlab Simulink solver flow chart . . . . . . 30
3.9 Validation of the cyclic transient Matlab/Simulink solver
with ESATAN/ESARAD . . . . . . . . . . . . . . . . . . . . . 31
3.10 Hot Case definition, influence of orbit orientation . . . . . . 33
3.11 Cold Case temperature evolution . . . . . . . . . . . . . . . . 34
3.12 T
max
& T
min
sensitivity to aluminum panel thermo-optical
properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.13 Advanced simulink thermal model hot case results . . . . . 36
3.14 Advanced simulink thermal model cold case results . . . . . 36
4.1 Thermal model processing flow chart . . . . . . . . . . . . . 39
4.2 Geometric Mathematical Model nodal breakdown in ESARAD 41
viii
4.3 On the left, the CubeSat with one aluminum panel before
integration. On the right, the panel integrated with the cells
by EADS Astrium . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.4 Planet centered Local Orbit Coordinate System (LOCS),
from ESARAD User Manual . . . . . . . . . . . . . . . . . . . 43
4.5 MCRT convergence from ESARAD user manual [2] . . . . . 43
4.6 Radiative exchange factor convergence with number of rays
fired (MCRT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.7 Evaluation of the conductive link of aluminum frame
through a finite element analysis within SamcefField . . . . 45
4.8 Solar cell power balance . . . . . . . . . . . . . . . . . . . . . 46
4.9 Conductive link computation through a finite element anal-
ysis within SamcefField of a midplane standoff . . . . . . . 49
4.10 Hot case absorbed power . . . . . . . . . . . . . . . . . . . . 51
4.11 Hot case temperatures . . . . . . . . . . . . . . . . . . . . . . 51
4.12 Evolution of the temperatures during one orbit for a rota-
tion rate of 5 deg/s . . . . . . . . . . . . . . . . . . . . . . . . 52
4.13 Cold case absorbed power . . . . . . . . . . . . . . . . . . . . 53
4.14 Cold case temperatures . . . . . . . . . . . . . . . . . . . . . . 54
4.15 Influence of the EPS-BAT spacers conductivity on batteries’
temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.1 Battery test setup . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.2 Battery test Simscape
TM
model . . . . . . . . . . . . . . . . . . 58
5.3 Battery test comparison between measurements and ad-
justed model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.4 Evolution of the emissivity and absorptivity of aluminum
sample subjected to a conversion coating surface treatment
in function of the reaction time [61] . . . . . . . . . . . . . . 60
5.5 A schematic representation of the general thermographic
measurement situation from the ThermaCAM User Manual
[21]. 1: Surroundings; 2: Object; 3: Atmosphere; 4: Camera . 61
5.6 Influence of the object temperature and emissivity for a
thermographic measurement . . . . . . . . . . . . . . . . . . 62
5.7 aluminum frame emissivity test setup . . . . . . . . . . . . . 63
5.8 Infrared capture of the structure heated up to 90

C . . . . . 63
5.9 aluminum frame emissivity test results . . . . . . . . . . . . 64
5.10 Aluminum frame contacts in front of bright background.
On the left, face 6 and 4, fastened with 3 screws and on the
right, face 6 and 1, fastened with one screw . . . . . . . . . . 64
5.11 Infrared images Battery test setup . . . . . . . . . . . . . . . 65
6.1 Midplane standoffs fastening to the aluminum frame . . . . 68
6.2 Detailed Thermal Model nodal breakdown of the aluminum
frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
6.3 Geometric Mathematical Model in ESARAD . . . . . . . . . 71
6.4 TRW and Lockheed Martin Bolted-Joint Resistance Data [24] 71
6.5 Thermal conductance between two thin plates of different
section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.6 external layout and corresponding thermal conductance . . 73
6.7 On the left, inserted copper washers in EPS2 engineering
model. On the right, the inter-PCBs conductive model . . . 74
6.8 Effect of temperature and aging on solar cells efficiency . . 76
6.9 ESARAD radiative coupling check for one PCB node (left:
analytical, right: MCRT) . . . . . . . . . . . . . . . . . . . . . 78
6.10 check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.11 Checking the conductive links inside PCBs through
isotherm shape . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.12 Check of the conductive links through the bus and equiva-
lent resistance model . . . . . . . . . . . . . . . . . . . . . . . 80
6.13 Cold case absorbed power . . . . . . . . . . . . . . . . . . . . 81
6.14 Hot case absorbed power . . . . . . . . . . . . . . . . . . . . 81
6.15 Cold case absorbed power . . . . . . . . . . . . . . . . . . . . 82
6.16 Hot case 1 heat flow map . . . . . . . . . . . . . . . . . . . . 83
6.17 Hot case 1: Hot spot due to the dissipation transistor and
temperature distribution on the antennas . . . . . . . . . . . 84
6.18 Hot case 2 heat flow map . . . . . . . . . . . . . . . . . . . . 85
6.19 Hot case 2: Hot spots due to the dissipation transistor and
amplifier on COM PCB . . . . . . . . . . . . . . . . . . . . . . 86
6.20 Battery temperature profile of a typical LiPo battery, here
the KOKAM SLPB723870H4 [32] . . . . . . . . . . . . . . . . 87
6.21 Heating system control flow chart . . . . . . . . . . . . . . . 88
6.22 Batteries’ PCB fastening concept . . . . . . . . . . . . . . . . 89
6.23 Equivalent conductivity of BAT PCB fastening in function
of the PA washer thickness and thermal conductivity of the
spacer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.24 Effect of insulating the batteries on the heater required
power for a given threshold [5

C] . . . . . . . . . . . . . . . 90
6.25 Evolution of the battery minimal temperature, batteries’ ca-
pacity decrease (worst voltage U=2.5 V) and time the heater
is turned on in function of the heater power for different
thresholds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.26 Temperature of a SLB 603870H KOKAM battery during the
cold test [34] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.27 Power distribution in the equivalent resistance and the tran-
sistor of the dissipation system . . . . . . . . . . . . . . . . . 94
6.28 Thermal strap (angle bracket in red) design on engineering
EPS PCB (left) and its fastening configuration with the re-
located resistances on antenna panel (CATIA Model on the
right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.29 New design hot case heat flow map . . . . . . . . . . . . . . 97
6.30 Amplifier temperature increase in function of its relative lo-
cation on the COM PCB . . . . . . . . . . . . . . . . . . . . . 98
7.1 Thermal margin terminology for NASA/JPL programs [24] 104
7.2 Thermal margin terminology from ECSS-E-10-03A standard
[17] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
7.3 Thermal vacuum cycling test sequence [17] . . . . . . . . . . 105
7.4 Example of thermal shroud used for the testing of Planck in
FOCAL 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
7.5 Thermal vacuum cycling test set-up . . . . . . . . . . . . . . 110
6 Conductance model of face 3 . . . . . . . . . . . . . . . . . . 117
7 Face 3 conduction SamcefField results . . . . . . . . . . . . . 118
8 Conductance model of faces 1, 2 and 5 . . . . . . . . . . . . . 119
9 View factor between two parallel rectangles [15] . . . . . . . 120
10 Advanced Simulink thermal model main high level diagram 121
11 Advanced Simulink thermal model low level diagram:
global model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
12 Advanced Simulink thermal model low level diagram: PCB
stack model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
13 Equivalent conductivity of BAT PCB fastening in function
of the PA washer thickness and thermal conductivity of the
spacer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
14 Equivalent conductivity of BAT PCB fastening in function
of the PA washer thickness and thermal conductivity of the
spacer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
List of Tables
2.1 Thermal control means . . . . . . . . . . . . . . . . . . . . . . 17
3.1 OUFTI-1’s Thermal Requirements . . . . . . . . . . . . . . . 21
3.2 OUFTI-1s thermal heat capacity . . . . . . . . . . . . . . . . 31
3.3 Preliminary Hot & Cold case definition . . . . . . . . . . . . 32
3.4 External area ratios . . . . . . . . . . . . . . . . . . . . . . . . 34
3.5 Sensitivity coefficients to thermo optical properties . . . . . 35
4.1 Thermo Optical Properties . . . . . . . . . . . . . . . . . . . . 42
4.2 Equivalent conductance between aluminum frame faces . . 45
4.3 aluminum alloys properties . . . . . . . . . . . . . . . . . . . 45
4.4 Nodal capacitances . . . . . . . . . . . . . . . . . . . . . . . . 46
4.5 Thermal conductivity of Kapton
TM
[14] and RTV S691 adhe-
sive [18] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.6 PCBs links materials thermal conductivities . . . . . . . . . . 48
4.7 PCBs links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6.1 28% solar cells efficiency parameters [7], for C
S
=
1367 [W/m
2
] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.2 DTM Hot Cases internal dissipation . . . . . . . . . . . . . . 77
6.3 Effects of the different solutions to the hot case issues . . . . 96
6.4 Comparison of hot case temperatures before and after mod-
ifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.1 Legend and symbols for thermal vacuum (and cycling) test
sequences (Fig. 7.3) [17] . . . . . . . . . . . . . . . . . . . . . 106
7.2 Thermal cycling test parameters (qualification) (Fig. 7.3) [17] 106
7.3 Qualification and acceptance test levels and durations ac-
cording to ECSS-E-10-03A standard [17] . . . . . . . . . . . . 106
7.4 OUFTI-1’s test temperatures definition . . . . . . . . . . . . 107
7.5 Thermal testing heating/cooling methods [24] . . . . . . . . 109
6 Aluminium frame faces conductances, in [W/K] 10
2
. . 119
xii
Thesis outline
This thesis focuses on the thermal analysis of the nanosatellite OUFTI-
1, developed at the University of Liège. Through it, the thermal environ-
ment the satellite will be exposed to is studied. The ultimate goal of this
thesis is to take suitable measures to ensure all the components will be
functioning in their safe range of temperatures and a proper heat rejec-
tion.
This work will be divided into 7 parts : first, a brief introduction
about the project, the mission and the CubeSat concept. Then a prelim-
inary analysis is conducted through Matlab/Simulink softwares to have a
first guess of the temperatures the CubeSat will undergo. After that, two
ESATAN/ESARAD are developed: a first one to confirm the results of the
single node Matlab/Simulink model and a second one, far more detailed,
to reveal and/or confirm the possible issues. With this detailed model,
a series of parametric analysis are performed to enable suitable design
modification accordingly with the other subsystems to enforce thermal re-
liability with sufficient margins. Before developing this detailed model,
measurements will be performed on critical elements such as the battery
or the main structure. Finally, thermal vacuum and cycling tests that
will be performed next year at the Centre Spatial de Liège will introduced.
Then we will draw conclusions and propose future development.
1
1
OUFTI-1 project, mission and
nanosatellite
1.1 The Project Story
The LEODIUM program
The OUFTI-1 project takes place within the framework of a long-term
goal program called LEODIUM. LEODIUM, besides meaning Liège in
Latin, stands also for Lancement En Orbite de Démonstations Innovantes d’une
Université Multidisciplinaire (Launch into Orbit of Innovative Demonstrations
of a Multidisciplinary University). It involves both the University of Liège
and Liège Espace, a consortium of space industries and research centers
in the Liège region. In 2005, Mr. Pierre Rochus, as the president of Liège
Espace and Deputy General Manager for Space Instrumentation of the
Centre Spatial de Liège, was charged with the training of students to the
design of miniaturized satellites. Different possibilities were foreseen and
the project finally began with the participation in the Student Space Ex-
ploration and Technology Initiative (SSETI) of the European Space Agency.
For two years, students were involved in the European Student Earth Or-
biter (ESEO) with the design of the solar panels deployment system and
also in the development of the the Narrow Angle Camera (NAC) for the
European Student Moon Orbiter (ESMO).
A CubeSat, why not ?
The idea of developing a nanosatellite was already in the mind of many
people at ULg but yet nothing concrete had began. One must wait until
September 2007 when Mr. Luc Halbach, sales manager of Spacebel, pro-
posed to test a new amateur radio digital technology in space on board
of a CubeSat : the D-STAR protocol. It did not even take one month for
a team of students and professors to set up around the newborn project.
Yet it needed a name : OUFTI-1, which is a typical expression of the city
of Liège, standing for Orbiting Utility For Telecommunication Innovation. In
fact, the University of Liège, through the LEODIUM program, has now the
ambitious goal to develop other CubeSats to keep on giving students satel-
lite hands-on experience. Last year, in their master theses, Stefania Galli
and Jonathan Pisane performed respectively the feasibility study and a
detailed analysis of the D-STAR protocol.
2
Chapter 1. OUFTI-1 project, mission and nanosatellite
Vega Maiden Flight
In October 2007, the ESA Education Office in cooperation with the Di-
rectorate of Legal Affairs and External Relations and the Vega Programme
Office in the Directorate of Launchers, issued a first Announcement Op-
portunity offering a free launch on the Vega maiden flight. After present-
ing the project at the Vega Maiden Flight CubeSat Workshop at the Eu-
ropean Space Research and Technology Center (ESTEC) in January 2008,
ESA published a call for proposal for CubeSat on board of Vega to which
we submitted our own proposal in March 2008. We finally got a positive
answer in June 2008, thanks to the good presentation made 6 months ear-
lier. Eight other CubeSats are selected (Xatcobeo for Vigo/INTA in Spain,
Robusta for Montpellier in France, UNICubeSat for Rome, AtmoCube for
Trieste and e-st@r for Turin in Italia, PW-Sat for Warsaw in Poland, Go-
liat for Bucharest in Roumania and SwissCube for Lausanne in Switzer-
land) and there are also two back-up (HiNCube for Narvik in Norway
and UWE-3 for Würzburg in Germany). In April 2008, Amandine Denis
presented the project at the 5
th
Annual CubeSat Developers’ Workshop in
San Luis Obispo, USA. This year again, the status of our project has been
presented during the 2
nd
European CubeSat Workshop in January 2009 at
ESTEC and at the 6
th
Annual CubeSat Developers’ Workshop in San Luis
Obispo, USA.
Initially, the Vega Maiden Flight was scheduled for November 2009.
This left us only a little bit more than one year to develop our CubeSat.
This tight schedule involved some decisions that will be described later.
Nevertheless, the flight has recently been delayed to spring 2010.
The Team
The initial team was composed of 2 second master students (Stefania
Galli and Jonathan Pisane), 1 first master student (Philippe Ledent) two
graduate students (Amandine Denis and Jean-François Vandenrijt), three
professors (Prof. Gaëtan Kerschen (A&M Dept.), Prof. Pierre Rochus (CSL
and A&M Dept.), Prof. Jacques Verly (EECS Dept.)) and Luc Halbach from
Spacebel.
This year, more students joined the project so that we are now 13 stu-
dents divided up into 9 subsystems:
ADCS : Samuel Hannay
1
COM : Renaud Henrard
2
and Francois Mahy
3
EPS : Philippe Ledent
3
and Pierre Thirion
3
GND : Laurent Chiarello
4
MECH : Jérôme Wertz
5
MIAS : Vincent Beukelaers
1
OBC : Nicolas Evrard
6
, Johan Hardy
6
and Damien Teney
7
STRU : Gauthier Pierlot
1
THER : Lionel Jacques
1
1
ULg, Aerospace engineering;
2
Isil Institute, Electrical engineering;
3
ULg, Electrical
engineering;
4
ULg, Computer engineering;
5
Gramme Institute, Mechanical engineering;
6
Gramme Institute, Electrical engineering;
7
ULg, Computer science.
Lionel Jacques
2
nd
Master in Aerospace Engineering
3 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 1. OUFTI-1 project, mission and nanosatellite
Management is carried out by Amandine Denis and Jonathan Pisane,
both PhD students at the University of Liège.
1.2 The Mission
In addition to give students satellite hands-on experience, OUFTI-1
mission has three main objectives. OUFTI-1 will be the first satellite ever
equipped with D-STAR. The primary goal is thus the space qualification
of this new technology so that a functional D-STAR repeater can be given
to the amateur radio community. Hence, we will use the amateur radio
frequency bandwidth reserved for the communications between the satel-
lite and the ground station. Furthermore, seeing the worldwide coverage
of the amateur radio community, they will reveal to be an invaluable help
us to keep our nanosatellite in good health when it is not in sight of our
ground station.
But what is D-STAR? As previously said, it is a new amateur-radio
communication protocol. The main difference between conventional pro-
tocols and D-STAR is that it is digital. The quality on the voice is then
definitely better. Another D-STAR key feature is that it allows simultane-
ous data and voice transmission.
What becomes D-STAR in space? In fact, using D-STAR in space have
some consequences. The frequency shift due to Doppler Effect during one
pass of the satellite is indeed too large for the acceptable bandwidth of the
ICOM 2820 transceiver Doppler Effect compensation capabilities. Unfor-
tunately, this transceiver is, at the moment, the only one able to deal with
D-STAR protocol available on the market. Doppler Effect will so have to
be on-board compensated. Seeing that, there will be two system-selected
doppler-compensated coverage zones : the first around ULg for control
and another one dynamically determined. Concerning the second one,
there will probably be a registration procedure on our website for the user
to reserved one pass of the satellite over a specific region. This is illustrated
on the figure 1.1. This zone definition fits well with an emergency usage
of D-STAR in space : in case of large scale disasters (Katrina Hurricane or
more recently the earthquake that hits center Italy), common communica-
tion networks are often unavailable. D-STAR is then very useful because
this recently developed protocol allows not only digital voice communi-
cation but also data transmission such as GPS data at the same time! The
disaster zone could then be selected to be Doppler compensated and allow
amateur radio to use this D-STAR satellite repeater to communicate and
send GPS data.
Anyway, as it is also shown on the figure 1.1, any user who has the
ability to make his own Doppler compensation could technically commu-
nicate with the CubeSat when it is in sight from the user and in D-STAR
mode (and not in AX.25 mode for Telecommand/Telemetry communica-
tion).
As previously said, OUFTI-1 has two other payloads : high efficiency
solar cells and an innovative EPS. As it will be presented in the next para-
graph, nanosatellite are ideal low cost solutions for testing new technolo-
gies. Concerning OUFTI-1, AZUR SPACE Solar Power GmbH proposed
us to test their new solar cells. These are 30% efficiency triple junction
Lionel Jacques
2
nd
Master in Aerospace Engineering
4 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 1. OUFTI-1 project, mission and nanosatellite
• 2 system-selected doppler-
compensated coverage zones
- ULg for control
- Dynamically
determined
• Personnal doppler –
compensation possible
within OUFTI-1 coverage
zone
Figure 1.1 – Mission description : Doppler-Effect compensated zones
GaAs cells, compared to their previous 28% model or their silicon cells
(type S-32) having only an efficiency of 17%. The cells are thus going to be
tested in real conditions. AZUR SPACE is a world recognized company in
space solar cells production and nowadays, more than 300 satellites have
been equipped with solar cells without any failure[7].
The third payload is a new digital electrical power unit developed in
cooperation with Thales Alenia Space (ETCA) about which we will come
back later. This innovative EPS is digitally controlled and based on a PIC
microcontroller and other components such as planar transformers. When
the batteries voltage is high enough (and the CubeSat normally works),
digital EPS will supply and be connected to the 3.3V power bus. Yet, this
EPS has a relatively lower efficiency than the analog 3.3V converter (50%
compared to 90%).
1.3 The Satellite
Up to this point, even if the word CubeSat has been used at least ten
times, it has not yet been presented. But what really is a CubeSat or a
nanosatellite ? In fact, for several years satellite have become larger and
larger (e.g. Envisat launched in 2002 which weights more than eight tons
and is 26 meters tall or Hubble in 1990 with its 11 tons and 16 meters)
but such missions required generally about ten years of development and
billions of euros. This is why there is a new tendency in reducing the size
and thus costs and development time. Actually, there is now a classifi-
cation of miniaturized satellites according to their wet mass : minisatel-
lites (between 100 and 500 kg), microsatellites (between 10 and 100 kg),
nanosatellites (between 1 and 10 kg) and finaly Picosatellites (between
0.1 and 1 kg). During the Second European Workshop at ESTEC, Prof.
Lionel Jacques
2
nd
Master in Aerospace Engineering
5 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 1. OUFTI-1 project, mission and nanosatellite
Figure 1.2 – OUFTI-1’s three payloads
Bob Twiggs from the Department of Aeronautics & Astronautics at Stanford
University presented the concept of even smaller satellites : femtosatel-
lites, weighting less than 100g, for teaching purposes ! Miniaturized and
nanosatellites are now very sought-after by industrials and academics for
testing new technologies under real conditions at low cost and short de-
velopment time.
Within this framework, Prof. Bob Twiggs originally proposed the
10x10x10cm 1kg CubeSat Standard. Through this standard definition, he
wanted to promote an easier space access and hands-on university-level
space education. Standardization of the bus allows standardization of the
interface between the satellite and the launcher so that standard deploy-
ment systems can be developed and launch opportunities increased. As an
educational university program, the CubeSat concept is based on lowering
the costs and simplicity. The use of non-space rated Commercial-Off-The-
Shelf (COTS) components is thus unavoidable. The CubeSat leitmotiv is
then :
In 2000, Pumpkin, Inc. decided to design a reliable off-the-shelf Cube-
Sat Kit conform to CubeSat specifications to facilitate CubeSat develop-
ment within tight schedules. As said in the presentation made Andrew
E. Kalman, Pumpkin’s president, the CubeSat Kit is strong, modular, light,
scalable, customizable and affordable [47]. Now, Pumpkin offers 1-unit (1U),
2-unit and 3-unit structures and also OBC, EPS and ADCS modules. In
conjunction with Prof. Jordi Puig-Suari at California Polytechnic Univer-
sity San Luis Obispo, a Picosatellite Orbital Deployer (POD) has been de-
veloped : the P-POD (Poly-Picosatellite Orbital Deployer). It can launch
up to 3 single-unit CubeSats or any fitting combination. Nevertheless, one
had to waid until 2003 to see first CubeSats in space.
Lionel Jacques
2
nd
Master in Aerospace Engineering
6 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 1. OUFTI-1 project, mission and nanosatellite
Figure 1.3 – On the left, the CubeSat Kit Structure, on the right, the P-POD
1.4 The Subsystems
ADCS [25]
Figure 1.4 – Earth’s magnetic
field orientation on the orbit[25]
Since our payloads do not require a pre-
cise attitude control, our CubeSat will be
passively controlled with hysteretic mate-
rials and a permanent magnet. The initial
rotation rate given at the P-POD ejection
is variable. This explains the main goal of
hysteretic materials: to prevent from too
high rotation rates which could cause an
additional unfavorable modulation of the
communication signal. Hysteretic materi-
als will thus damp the rotational energy
acquired at the deployment. But perturba-
tion torques (gravity gradient, solar pres-
sure, atmospheric drag...) are such as, even
with hysteretic materials, the CubeSat will
never stop rotating.
On another hand, the permanent mag-
net will align the satellite on Earth’s magnetic field which is represented
during one orbit on the figure 1.4. As the lines of the magnetic field are al-
ready falling above Belgium, the magnet will be perpendicular to the face
with the antennas so that antennas are nearly perpendicular with zenith
above Belgium. This is favorable by seeing antennas’ radiation pattern.
Lionel Jacques
2
nd
Master in Aerospace Engineering
7 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 1. OUFTI-1 project, mission and nanosatellite
COM [36],[28]
Figure 1.5 – OUFTI-1’s test RX
PCB[36],[28]
OUFTI-1 will use the amateur radio com-
munication bands : 435 MHz for the up-
link and 145 MHz for the downlink. As de-
scribed previously, D-STAR communica-
tion protocol will be used as payload. Nev-
ertheless, AX.25 protocol has been chosen
for TC/TM
1
to avoid TC/TM relying on
the experimental communication payload.
Anyway, there is also a reliable CW bea-
con sending continuously satellite’s key
housekeeping parameters that any ham-
radio operator in the world will be able to
listen and forward to us.
EPS [55],[33]
Figure 1.6 – Main EPS
engineering model
The primary function of the EPS is to store
the energy collected by the solar cells in
the batteries and provide the bus with the
required voltages. Concerning OUFTI-1, 5
faces of the CubeSat will be covered with 2
solar cells connected in series, the remain-
ing face being dedicated to the antenna de-
ployment mechanism. As previously pre-
sented, solar cells are our second payload
and are provided AZUR SPACE. The ex-
act battery model is not yet defined but
Varta or Kokam lithium polymer batteries
are nevertheless foreseen for their high en-
ergy density. The EPS contains three power bus: 3.3V, 5V and 7.2V.
In addition to the main and robust 3.3V converter, an innovative digital
EPS is being developed in cooperation with Thales Alenia Space ETCA,
already presented in a previous paragraph.
GND [10]
The Ground subsystem consist in the tracking antennas, its remote control
and the scheduler. The Main Tracking Station, located on the campus of
the University of Liège, will be made of two separated units : the first one
located at the Poste Central de Commande (PCC) of the university and
the second one, the User Tracking Station (UTS), located at the Montefiore
Institute. The first one can only deal with D-STAR transmission while
the UTS also supports communication with the CW beacon. The GENSO
2
compatibility is thus foreseen for the UTS. A Backup Tracking Station will
1
TC : Telecommand, TM : Telemetry
2
GENSO states for Global Educational Network for Satellite Operations. GENSO is a
worldwide network of ground stations and spacecraft interacting via a software standard.
Through this, GENSO aims to increase the return from educational space missions.
Lionel Jacques
2
nd
Master in Aerospace Engineering
8 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 1. OUFTI-1 project, mission and nanosatellite
also be installed at the Redu Euro Space Center. It will be fully redundant
and linked by a high speed internet tunnel with the first one.
MECH [63]
Figure 1.7 – OUFTI-1’s antenna
deployment mechanism panel[63]
Using two frequencies (uplink and down-
link), OUFTI-1 will have two monopole
λ
4
antennas : a 17cm and a 50cm one.
Throughout the launch sequence, anten-
nas will have to be folded since the Cube-
Sat will be in the P-POD until deployment.
Therefore, the MECH subsystem has to de-
sign a reliable deployment mechanism. As
shown on the figure 1.7, antennas will be
winded around a dedicated panel. During
the launch phase, they will be maintained
by a small thread which will then be heated up to its melting point to
release both antennas.
MIAS [9]
Mission Analysis consists in analyzing the global comportment of the
satellite in orbit. This includes lifetime evaluation, link budget with the
ground station calculation, Doppler estimation and so on. . . To do this, a
global simulator is being developed in Matlab Simulink environment in
correlation with STK software. The simulator include orbit propagation,
taking into account Earth’s oblateness, atmospheric drag, solar radiation
pressure. . . A simple model of each subsystem is also embedded : EPS for
power calculation, ADCS for the attitude determination, COM for the link
budget and Doppler considerations. A thermal model will also be pro-
vided and described later. A screenshot of the simulator is shown on the
figure below.
Figure 1.8 – Snapshot of the Simulator developed by V. Beukelaers in Simulink[9]
Lionel Jacques
2
nd
Master in Aerospace Engineering
9 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 1. OUFTI-1 project, mission and nanosatellite
OBC [54],[19],[26]
Figure 1.9 – OUFTI-1’s second
OBC PCB[19]
The On-Board Computer is the brain of the
CubeSat : it controls data flow, deals with
the different operational modes, ensure
telemetry data formatting and storage and
telecommand data decoding and manage-
ment. OUFTI-1 will have two fully redun-
dant OBC : a first robust one provided in
the Pumpkin’s CubeSat Kit (FM430 flight
module with Texas Instruments single-
chip 16-bit MSP430). But this one contains
many unnecessary functionalities for us.
This is why a second lighter OBC has been
developed (OBC2). If this home proves it-
self, it will probably be used as the main and only OBC for the next
OUFTI Missions. The picture 1.9 shows the home made OBC Printed Cir-
cuit Board (PCB). OBC also deals with housekeeping parameters such as
batteries’ voltage, solar cells’ currents, temperatures. . .
STRU [42]
Figure 1.10 – Finite element
modal analysis with
SamcefField[42]
The first key objective of the Structure and
Configuration subsystem is to position the
many components in order to be compli-
ant with the CubeSat design specification
and many other constraints (thermal, ra-
diations, high frequency issues). The sec-
ond objective is to ensure withstanding to
the harsh launch environment : even in
the P-POD, the nanosatellite undergoes ex-
tremely strong vibrations. Therefore, both
random and sine vibrations tests are per-
formed in correlation with Samcef finite
element models. The CubeSat should ulti-
mately proof that its structure and equip-
ments are capable of withstanding the maximum expected launch envi-
ronment through a series of acceptance and qualification tests. Those final
tests will be performed at the Centre Spatial de Liège.
Seeing the tight schedule, the idea to buy the CubeSat Kit structure and
OBC module (FM430) was accepted to save time and let us concentrate on
our payloads and other subsystems.
Lionel Jacques
2
nd
Master in Aerospace Engineering
10 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 1. OUFTI-1 project, mission and nanosatellite
1.5 OUFTI-1’s anatomy
Throughout this work, different components and specific parts of
OUFTI-1 will be referred to. These components are therefore introduced
here through relatively detailed exploded view (figure 1.11).
The skeleton of the satellite consist of Pumpkin’s 1U structure itself
divided into three parts : the lateral frame and the top and bottom faces.
The last two are fixed on the main frame with M3 screws not represented
here (4 screws for the up face and 6 for the bottom one). We called the
ports or rear side of the CubeSat the side where USB, remove-before-flight
pin hole and other ports are located.
The solar cells will be glued on aluminum panels. The panels must be
such as they keep the rails free (at least 8.5mm). By the rails, one means
the four vertical edges of the lateral frame which are hard anodized be-
cause they are the only surfaces in contact with the P-POD. Up to now, all
aluminum panels will also be glued on the structure to avoid remachin-
ing. As depicted on the view, only five of the six faces of the cube will be
covered with solar cells. The remaining one is dedicated to the antenna
deployment mechanism panel. OUFTI-1 will indeed have two antennas
described in a later paragraph.
Inside OUFTI-1, there are five printed circuit boards (PCBs). From bot-
tom to top, one has the OBC (CubeSat Kit’s FM430 Flight Module) fol-
lowed by its redundant home-made counterpart, the OBC2. Above them
lies the EPS also followed by the payload : the innovative EPS (EPS2).
The last PCB is the communication one on which will be located ampli-
fiers, quartz and other electronics circuits. While the OBC is the CubeSat’s
brain, its heart is the EPS with its two LiPo batteries fixed on a secondary
PCB and is strategically positioned as the central PCB. This PCB layout
has been driven by many constraints and considerations among these the
CubeSat specification stating that the center of gravity of the CubeSat has
to remain inside a 4cm diameter sphere centered on the geometric center
of the cube.
All the PCBs are maintained together thanks to four endless screws.
They are fixed at their bottom on the bottom face of the frame and at
their top on the lateral frame with the "midplane standoffs". The separation
between PCBs is ensure by spaces which are small aluminum cylinders.
We already sees that there are only 8 contact points between the PCB stack
and the external structure relies : four with the bottom face and 2 2 with
the lateral frame.
Lionel Jacques
2
nd
Master in Aerospace Engineering
11 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 1. OUFTI-1 project, mission and nanosatellite

COM
EPS2
EPS1
OBC2
OBC1
Aluminium
Panel
Lateral aluminium
frame
Solar Cell
Spacer
Batteries
Ports side
Bottom aluminium
frame
Top aluminium
frame



Antenna deployment
mechanism panel
Midplane
Standoff
Feet
Battery
PCB
Figure 1.11 – OUFTI-1 exploded view
Lionel Jacques
2
nd
Master in Aerospace Engineering
12 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
2
The Thermal Control
Subsystem
This chapter introduces the basics of spacecraft thermal control. It be-
gins with a brief recall of heat transfer and then thermal environment will
be exposed. After that, we will describe in a few words what is the aim of
the thermal control subsystem : why and how dealing with space thermal
issues.
2.1 Space heat transfer
In general, there are three main heat transfer modes
1
: conduction,
radiation and convection.
2.1.1 Conduction
In space, due to the extremely low residual pressure, only conduction
and radiation modes are present
2
. Conduction is governed by Fourier’s
Law. For an isotropic material:
q = k

T
where k is the constant thermal conductivity and where q is the heat
flux. This equation can be rewritten in the case of a steady unilateral flow
through a surface of a thickness L and constant area S as follows :
Q = qS =
kS
L
∆T = GL
(i,j)
(T
i
T
j
)
This defines the thermal conductance GL
(i,j)
between the isothermal
surfaces i and j.
2.1.2 Radiation
On the other hand, radiation heat transfer is governed by Stefan-
Boltzmann’s Law stating that the black-body irradiance is proportional
to the fourth power of its temperature :
1
Ablation, a combination of these three processes with chemical reaction, is also con-
sidered for atmospheric re-entry vehicles
2
Nevertheless, convection must be taken into account for manned mission, launch-
ers. . . (ISS, Shuttle,. . . )
13
Chapter 2. The Thermal Control Subsystem
E
BB
(T) =
_

0
E
BB
(λ, T)dλ = σ
0
T
4
[W/m
2
]
where σ
0
is the Stefan-Boltzmann’s constant in vacuum defined in
function of the universal constants : π, Boltzmann’s constant k, the speed
of light in vacuum c
0
and Planck’s constant h :
σ =

2
k
4
15c
2
h
3
= 5.67051 10
8
[Wm
2
K
4
]
E
BB
(λ, T) is the hemispherical spectral emissive power of a black-body
and is given by Planck’s Law :
E
BB
(λ, T) =
2πhc
2
λ
5
(e
hc
kλT
1)
[W/(m
2
.µm)]
Figure 2.1 – Planck’s black body radiation curves for different temperatures
Planck’s law is illustrated in the figure 2.1. For a given temperature,
there is a maximum in the energy distribution. The corresponding wave-
length is governed by the Wien’s displacement law describing the location
of the maximum in the hemispherical spectral emissive power :
λ
max
=
b
T
[m]
with b = 2.897 10
3
[m K].
All these formulas concerned the black-body which is idealized object
absorbing all radiant energy from any direction or wavelength and emit-
ting in any direction isotropically. The radiated energy of the black-body
only depends on its temperature. But a real body can absorb, reflect and
transmit radiation energy so that absorptivity α, transmittivity τ and re-
flectivity ρ quantities are defined, all wavelength and angular dependent.
For a semi-transparent body, energy conservation leads to the following
equality
α(θ, λ) + ρ(θ, λ) + τ(θ, λ) = 1
Lionel Jacques
2
nd
Master in Aerospace Engineering
14 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 2. The Thermal Control Subsystem
where θ is the angle of incidence. As there is no perfect black-body
in practice, the emissivity (λ) is defined as the ratio between the energy
emitted by a surface to that of a black body at the same temperature.
Absorptivity and emissivity can either be hemispherical or directional and
either total or spectral. The second Kirchhoff’s law states that for a given
direction θ, directional spectral absorptivity and emissivity are equal :
(θ, λ) = α(θ, λ)
But, in general, this is not true with total hemispherical values mainly
because of their strong wavelength dependence. Both α ans varies with
the angle of incidence but they are assumed to follow the Lambert’s law
stating that directional absorptivity/emissivity is proportional to cos θ
(maximum for normal incident angles and null for tangential ones). The
reflectivity can be either diffuse or specular (ρ = ρ
d
+ ρ
s
). When diffuse, it
follows the Lambert’s law as the emissivity and absorptivity while specu-
lar reflection follows Descartes’ law according to which the reflexion angle
is equal the the incident angle.
In space heat transfer, thermal engineers made the following assump-
tion : thermo-optical properties are assumed to be constant in two spectral
regions :
- the infrared spectrum, from λ = 4.25µm to 40µm, corresponding to
temperatures between 70K and 700K.
- the visible spectrum, ranging from λ = 0.3µm to 2.5µm associated to
a temperature range going from 1150K to 10000K
As the temperature of a spacecraft lies in the 70K-700K range, the emit-
ted radiation is infrared. But the source of the main incident radiation is
the sun which can be considered as a blackbody emitting at 5776K. This
temperature lies in the visible spectrum, as it is also shown by the figure
2.1. Actually, ESA and NASA thermal engineers adopted the following
convention : they call the (constant) emissivity (and absorptivity) in in-
frared wavelengths and α the (constant) absorptivity (and emissivity) in
visible wavelengths. This convention will also be adopted in this work.
Finally, surface thermo-optical properties are defined through eight
coefficients (6 independent) related by the two following equations :
+ ρ
IR,d
+ ρ
IR,s
+ τ
IR
= 1 α + ρ
VIS,d
+ ρ
VIS,s
+ τ
VIS
= 1
Where τ is the transitivity An important concept for radiative exchange is
the view factor F
(i,j)
between two surfaces S
i
and S
j
. It is the proportion of
all the radiative power leaving S
i
which directly strikes surface S
j
(without
any reflection) and only depends on the geometrical configuration :
F
i,j
:=
P
ij
P
i
=
1
πS
i
_
S
i
_
S
j
cos θ
i
cos θ
j
r
2
ij
dS
i
dS
j
Another useful concept is the Gebhart factor B
ij
: this factor takes into
account multiple reflections and represents the portion of the radiation
emitted by a surface S
i
and finally incoming to the surface S
j
. The Geb-
hart factor does not only depends on geometry but also on thermo-optical
Lionel Jacques
2
nd
Master in Aerospace Engineering
15 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 2. The Thermal Control Subsystem
properties. It is defined through the following equation, valid for diffuse
reflection
1
:
B
ij
= F
ij

j
+

k
F
ik
(1
k
)B
kj
The radiative coupling between two surfaces is directly deduced from
the Gebhart factor. Indeed, the radiative power Q
ij
flowing from S
i
to S
j
is equal to the difference between the power emitted by S
i
, absorbed by S
j
(
i
S
i
B
ij
σT
4
i
) and the one emitted by S
j
and absorbed by S
i
(
j
S
j
B
ji
σT
4
i
). At
equilibrium, Q
ij
= 0 and thus
i
S
i
B
ij
=
j
S
j
B
ji
. Out of equilibrium :
Q
ij
=
i
S
i
B
ij
σT
4
i

j
S
j
B
ji
σT
4
i
=
i
S
i
B
ij
σ(T
4
i
T
4
j
) = GR
(i,j)
σ(T
4
i
T
4
j
)
This defines the radiative exchange factor GR
(i,j)
between two surfaces
i and j.
2.2 The thermal environment
A satellite orbiting Earth has several heat sources, as represented on
the figure 2.2 :
1. Direct solar flux depending on sun distance, with a mean value
around 1367 [W/m
2
] at 1AU (1414 [W/m
2
] at winter solstice and
1322 [W/m
2
] at summer solstice [24]).
2. Albedo planetary reflected radiation. For Earth, the mean reflectivity
is assumed to be near 30%. But it can vary localy up to 40 or 80%
above shiny clouds and from 5 to 10% for ocean and forests[44].
3. Earth infrared radiation. Earth can be modeled as an equivalent
black-body emitting at 255 K [50], [44], [24].
4. Internal dissipated power in electronic components (Joule effect).
5. Not represented on the figure, aerothermal flux have also to be con-
sidered during launch or re-entry phases.
During the eclipse, only two heat sources are still present : Earth’s
infrared and internal dissipation and the spacecraft will be cooler. The
temperatures of the satellite tend thus to vary in a cyclic way along the
orbit, rising in sunshine and dropping during eclipse. The sky, called Deep
Space, is the main source of cold and can be seen as a black body emitting
at 3K. This temperature represent the radiation of the stars, the galaxies
and the Cosmic Microwave Background which will be studied by ESA’s
new satellite Planck.
1
A theory has been developed to take partial specular reflection into account [50] but
will not be used in this work because the reflections encountered here are mainly diffuse
Lionel Jacques
2
nd
Master in Aerospace Engineering
16 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 2. The Thermal Control Subsystem

Direct Solar Flux
Albedo
Earth
Infrared
Emitted
radiation
Internal
dissipated
power
Figure 2.2 – Thermal environment of a satellite in LEO
2.3 Thus, why Thermal Control ? and how ?
The primary objective of thermal control is thus to maintain temper-
atures of onboard equipments within specified ranges, that will be later
introduced for OUFTI-1. It also must ensure that temperature gradients
(spatial consideration) are not too large and a good temperature stability
(temporal consideration). Temperature gradients are undesirable for op-
tics pointing and alignment while stability and thus narrow temperature
ranges are important for the sensitivity of detectors. Therefore, the critical
equipments usually are optical instruments and electronics components
(mainly the batteries). For precise pointing satellites, thermo-elastic cou-
pling involving unwanted vibrations also occurs at eclipse in and out : this
is known as jitter phenomenon.
Then, how to fulfill all these requirements since the spacecraft under-
goes temperature variation solely determined by its radiative exchanges
with environment. In fact, there are two main categories of thermal control
means : active and passive ones. This can be summarized in the following
non exhaustive table :
P
a
s
s
i
v
e
Radiation Conduction
- coating - structural materials
- MLI blanket - doubler, filler, adhesive
- radiator - washer, strap, bolt, stand-off
Latent heat & Ablation
- Thermal protection system
- Phase change materials
A
c
t
i
v
e
Heaters Heat pipes & fluid loops
- thermostat control - fixed/variable conductance
- electronic control - loop heat pipe
- ground control - mono/diphasic fluid
Peltier element Louvres
Table 2.1 – Thermal control means
Lionel Jacques
2
nd
Master in Aerospace Engineering
17 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 2. The Thermal Control Subsystem
Except for heaters, most of active means required too much power and
mass or complex mechanisms and moving parts or even fluid packaging
and this is not compatible with the CubeSat concept. Furthermore, even if
active thermal control means are able to cope with larger heat loads, they
are also generally less reliable than passive ones and CubeSats have usu-
ally low power consumption and dissipation. Nevertheless, if the temper-
ature of an equipment dropped under its specified minimal one, a simple
way of heating is the use of heaters. The advantage of heater is that it can
nearly be of any dimension and is thus very convenient.
For all these reasons, the main and easiest way of doing thermal control
for nanosatellites remains passive control or small heaters.
2.4 Thermal control systems of other CubeSats
This is indeed the means other CubeSat teams have used to do thermal
control. Some of them are described below:
SwissCube
SwissCube nanosatellite [53] is developed at EPFL in Lausanne, Switzer-
land. SwissCube’s mission is to study the nightglow phenomon. Swiss-
Cube has its own developed structure fabricated by CNC milling and wire
electrical discharge machining and uses an active attitude control with
magnetotorquers, sun sensors, gyroscopes and magnetometers to stabi-
lize the satellite. Concerning the thermal control subsystem, their simula-
tions led them to use a 500 mW (25 Ω) heater to keep the two batteries
within their allowable range of temperatures. They are packed in an alu-
minum box to prevent deformation due to the vacuum environment and
are fastened on a copper plate as the heater, all with epoxy resin. The cop-
per plate is used because of its high thermal conductivity to conduct the
generated heat to the batteries. The heater is thermostatically controlled
avoiding the use of the OBC and is switched on under the 0

C threshold.
SwissCube has not yet been launched but is ready for it.
Compass-1
Compass-1 [12] is developed by the students of the University of Applied
Sciences in Aachen. Its mission is to take picture of the Earth and it has
therefore a active attitude control as SwissCube. Their thermal model led
them to use 2.01 W (68 Ω) heater located at the top of the battery box
combined with LM75 temperature sensors. The heater is switched on once
the temperature of the batteries drops under 5

C.
Compass-1 has already been launched on April, 28
th
, 2008 by the In-
dian PSLV launcher. After the successful launch, Compass-1 entered in a
vicious circle: the heater was too often switched on and the battery power
drops drastically. This untimely switch on of the heater occurred because
the temperature to which was compared the threshold was based on the
minimal temperature given by the three sensor inside the battery box. The
threshold has then been adapted but Compass-1 had to be reset for other
reasons so that the nominal threshold had also been reset. Unfortunately,
Lionel Jacques
2
nd
Master in Aerospace Engineering
18 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 2. The Thermal Control Subsystem
they lost their two ground stations at that time but, with the help of the
amateur radio community, they managed to recover the satellite and suc-
cessfully changed into its nominal mode. Now, all goes well for Compass-1
and many pictures have already been taken.
Del -C
3
Delfi-C
3
[13] is the nanosatellite developed and has been launched with
Compass-1 in April 2008. In addition to be a 3-Unit CubeSat, on the con-
trary of both previous CubeSats, the particularity of Delfi-C
3
lies in the
fact that it do not have any battery but deployable solar arrays. As the bat-
tery is usually the more critical component, Delfi-C
3
has no active thermal
control and its thermal behavior relies only the thermo-optical properties
of the outer surfaces.
Lionel Jacques
2
nd
Master in Aerospace Engineering
19 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
3
Preliminary thermal
analysis
A first analysis had already been developed in the Phase A study by S.
Galli. That was a steady state model and it did not take into account the
effect of satellite’s thermal inertia. In this chapter, a single node transient
model of the CubeSat will be implemented in Matlab Simulink environ-
ment.
This chapter will begin with a description of the notations and conven-
tions that will be adopted not only in this preliminary analysis but also
in the next chapters. Then, we will study the thermal environment and
implement a general way to compute the different incoming fluxes. After
that, the cyclic single node transient model will be introduced and first
guesses of hot and cold case will be computed. A sensitivity analysis will
also be performed.
3.1 Notations
Here are described the notations and conventions we will use through-
out this work.
Concerning the reference frame, it will be the one used by the Structure
and Configuration subsystem. It was shown on the exploded view
1
(figure
1.11) : its center is the geometric center of the CubeSat. The +X direction
will be perpendicular to the face where are fixed the antennas and where
the access ports are located and oriented to the opposite face, the +Z
direction will be the direction perpendicular to PCBs stack, from bottom
with the OBC PCB to top with the COM PCB. Finally, the +Y direction is
such as it forms with the other ones a right handed coordinate system. We
will also use the following numbering convention :
+X = 1 & X = 4
+Y = 2 & Y = 5
+Z = 3 & Z = 6
Concerning thermo-optical properties, ESA and NASA convention will
also be adopted, as described previously :
α = α
vis
=
vis
1
Excepted that it is not located at the center of the CubeSat for the sake of clarity
20
Chapter 3. Preliminary thermal analysis
=
IR
= α
IR
Environmental fluxes notation will be as follows :
Q
S
Solar power
Q
A
Albedo power
Q
E
Earth infrared power
Q
I
Internal dissipated power
3.2 Thermal Requirements
As explained in the previous chapter, the aim of thermal control is to
ensure that all onboard units will be working within their own allowable
temperature range. All electronic components are not yet defined but the
use COTS components mainly involves -40. . . +85

C temperature ranges.
For instance, this is the case for Pumpkin’s FM430 Flight Module (OBC),
following its datasheet [48], and also for the PIC18F2331 microcontroller
used on the experimental EPS [37]. Concerning COM PCB integrated cir-
cuits, the ADF7021 transceiver has also the same wide temperature range
[5]. A more critical component was the oscillator because of its important
temperature sensitivity. To avoid this potential problem, the use of a TCXO
(Temperature Compensated Quartz Oscillator) has been foreseen. In this
case, common TCXOs have the same temperature range.
The CubeSat Kit’s, as all the previous parts, also operates in the -40

C
to +85

C industrial temperatures range [47]. Typical values for solar cells
operating temperature ranges are -100 to +100

C [62].
Finally, the most critical part of the satellite remains the batteries.
Lithium polymer batteries have been chosen for their high specific capac-
ity. The final model is not yet determined but, among the possible fore-
seen models (VARTA PoLiFlex
R
[58] and Kokam
TM
SLPB 554374H [31]), all
have the same specifications depending on whether it is in charge, from 0
to +45

C, or discharge, from -20 to +60

C.
OUFTI-1 thermal requirements are summarized in the table 3.1.
Component T
min
[

C] T
max
[

C] Note
Main structure -40 +85
Solar cells -100 +100
Electronics -40 +85
LiPo Batteries
0 45 charge
-20 60 discharge
Table 3.1 – OUFTI-1’s Thermal Requirements
Another thermal requirement concerns the thermal expansion of the
structure. The main CubeSat Kit’s structure is made of aluminum 5052H32
alloy and thus already satisfies the differential dilatation condition with
the P-POD, also made of aluminum alloy (7075-T73). Concerning the pan-
els supporting the solar cells, it was decided, in agreement with the Struc-
ture and Configuration subsystem, that aluminum would again be more
suitable than CFRP for the same reason. Panels will indeed be glued on
the main frame (for a reason already explained in a previous chapter), and
minimize the difference between the thermal expansion coefficients of the
Lionel Jacques
2
nd
Master in Aerospace Engineering
21 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
panels and the frame was a good way to reduce the shear stress inside the
glue.
3.3 Thermal environment related to OUFTI-1’s orbit
Now, the different incident fluxes exposed in the previous chapter will
be calculated. To do so, a simple Simulink Model will be developed in
order to integrate the equations of motions of the two-body problem and
compute the position of the satellite required for the fluxes calculation.
3.3.1 The orbit
An orbit is described by its classical orbital elements (COE aka Keple-
rian elements) : the semi-major axis a, the eccentricity e, the inclination i,
the right ascension of the ascending node Ω or RAAN, the argument of
perigee ω and the true anomaly ν.
Figure 3.1 – Classical orbital elements
The reference line for the RAAN is taken as the vernal equinox, i.e.
the intersection of the equatorial plane with the ecliptic at vernal equinox
pointing directly to the sun. OUFTI-1’s orbit has a perigee altitude of
354km, an apogee altitude of 1447km and an inclination of 71

. The two
remaining parameters, the RAAN and perigee argument ω, are still un-
known.
To compute the satellite position around the earth, a simple Simulink
two body model has been derived. Initial position and velocity computed
from the COE are given as input to the Simulink 6DoF Euler Angles inte-
grator in addition to the acceleration given by equation 3.1:
a =
µ
r
3
r (3.1)
Lionel Jacques
2
nd
Master in Aerospace Engineering
22 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
where µ is the standard gravitational parameter of the earth andr the
position vector of the satellite. An approximate attitude is given by the
mean of initial euler angles and rotation rate of the satellite. The rotation
rates of the satellite will remain constant as we consider that no moment
act on it. As output, the integrator gives the position and direct cosine
matrix (DCM) for the next time step. The DCM represents the current ori-
entation of the satellite compared to the inertial axes, i.e. it is the rotation
matrix to pass from the inertial coordinate frame to the body axes.
Then, at each time step, the model compute the incident heat flux on
each face of the satellite. This is done in the Albedo and IR calculation box
described in the next paragraphs.
3.3.2 Solar flux
As the sun distance is extremely large, the rays coming from the sun
are assumed to be parallel. Furthermore, seeing the cubic shape of the
satellite, there is no coupling between the faces, i.e. any ray that is reflected
by one face will never hit another face. Therefore, the incident flux on each
face is simply obtained by multiplying the solar constant C
S
with the scalar
product between the normal to the face and the sun direction.
q
S,i
= C
S
(n
i
r
sun
) [W/m
2
]
The incident solar flux on each face is thus directly determined by the
direct cosine matrix (DCM) and only depends on the solar constant C
S
and the satellite attitude.
But how to know whether the spacecraft is in sunlight? The same
method as the one implemented by V. Beukelaers in the Simulator was
used: the Line-Of-Sight (LOS) method, developed by Alfano in 1991, de-
scribed in [56]. This method is based on determining if the line between
the satellite and the sun intersects the Earth : if it does, there is no LOS
and the satellite is in eclipse and in all other cases it is in sunlight. The
advantage of this method is that it avoids trigonometric operations. For
further details, refer to [56] and [9].
The sun direction will be taken as a constant and equal tor
sun
= [1 0 0].
This means that if the inclination is 90

, the orbital plane will be perpen-
dicular to the sun rays if Ω = 90

and parallel for Ω = 0

.
3.3.3 Albedo flux
The albedo flux is relatively more complex to evaluate. Indeed, it de-
pends on many parameters such as satellite’s position from the subsolar
point, Earth view factor and many others. Many different expressions or
tables can be found in the literature to compute the albedo but they are
often not easy to use or incomplete. For that reason, we decided to com-
bine two references : a simple formula coming from the notes of "Contrôle
thermique des engins spatiaux"[1] by I. Alet and an abacus coming from the
book "Spacecraft Thermal Control Handbook" by D. Gilmore [24].
Here is the formula and the abacus
1
is represented on the figure 3.3 :
1
initially coming from Lockheed Martin
Lionel Jacques
2
nd
Master in Aerospace Engineering
23 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
Reshape
U( : )
Matrix
Multiply
Matrix
Multiply
Initial Conditions : COE
a
e
nu
Raan
omega
i
r0
v 0
Gravitationnal Model
x_ECI a_ECI
Equations of Motion
Integration
a
r0
v 0
pqr0
euler 0
DCM
Xe
Epoch
−C−
Albedo and IR calculation
r_sun_ECI
r_sat _ECI
Cs
DCM
a
albedo
IR
albedo_face
IR_face
Solar _face
Normal
omega
14
psi
13
phi
12
11
10
9
8
7
6
5
4
r0
3
q0
2
p0
1
Normal
6
Solar _face
5
IR_face
4
albedo _face
3
IR
2
albedo
1
Normalize
Normalize
sqrt
r_sat
DCM
rhos
fcn
r_sun_ECI
DCM
cosines
Normals 3_ECI
fcn
r_sat
r_sun
theta
fcn
Sun_ECI
Sat_ECI
sunlight
fcn
Earth Infrared and Albedo computation
Cs
a
Altitude
rhos
theta
Albedo
IR
Albedo_face
IR_face
alb_IR
Dot Product
Bias
u−6378
a
5
DCM
4
Cs
3
r_sat_ECI
2
r_sun_ECI
1
Figure 3.2 – Simulink two-body propagator for incident ux computation : main
window (top) and Albedo and IR calculation box (bottom)
Lionel Jacques
2
nd
Master in Aerospace Engineering
24 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
q
A
= C
S
a [cos(0.9 θ)]
1.5
F
E
[W/m
2
]
where C
S
is the solar constant, a the albedo reflectivity coefficient, θ
the "sun-earth-satellite" angle and F
E
the view factor between the face and
the earth. The 0.9 coefficient in the cosine means that even if the satellite
has just come above the shadow part of the Earth, reflected rays are still
hitting the satellite. When θ = 0, the satellite is at the subsolar point and
the albedo is maximum.
Figure 3.3 – Incident albedo irradiation on a surface element in Earth orbit [24]
The formula required to compute the view factor between the face
and the earth, F
E
which is quite difficult since it depends both on the
altitude and orientation of the face. But, even if this view factor can be
easily hand computed with the abacus, this is no more easy to implement
within Matlab. Therefore, all these curves were digitized and, to simplify
their use, fitted with this formula :
F
E
= r
2.1
_
sin
_
ρ
2
__
e
(3.2)
Lionel Jacques
2
nd
Master in Aerospace Engineering
25 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
with the exponent e described as a function of r:
e = 160.31r
6
+723.36r
5
1380r
4
+1394.6r
3
780.65r
2
+226.81r 21.232
where r =
R

R

+h
. The fitting approximation of the view factor F
E
is
represented on the figure 3.4, in comparison with the Lockheed Martin
abacus.
0 20 40 60 80 100 120 140 160 180
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
ρ [deg]
V
i
e
w

f
a
c
t
o
r

F
E

Altitude [km]
200
500
1000
2000
3000
4000
6000
10000
20000
Lockheed Martin’s Data
Approximation
Figure 3.4 – Fitting of the Gilmore abacus curves with the formula 3.2
The equation and the abacus show that even if the scalar product n
f ace

r
sat
is negative (i.e. ρ < 90 deg and the face does not point directly toward
the Earth), the view factor is not null and the face still absorbs albedo or
infrared flux. This can be explained by the fact that the incoming rays are
no more parallel as it was the case for the sun rays since the satellite is
close to the emitting body that can no more be considered as a point. This
feature will involve a greater effective surface as it will be shown.
Of course, the combined model has been correlated with the abacus
method. This albedo model does not take into account local reflectivity
variation (clouds, forests, oceans. . . )
3.3.4 Earth infrared flux
Earth’s infrared flux, also depending on the same F
E
is now easier to
compute since F
E
is already known. By neglecting diurnal and seasonal
variation in the temperatures of the earth and considering it as an equiv-
alent 255K black body (thus emitting diffusely in all directions), one can
obtain the following equation :
q
E
= E
BB

T=255K
F
E
= 240 F
E
[W/m
2
]
Lionel Jacques
2
nd
Master in Aerospace Engineering
26 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
3.3.5 Aerothermal flux
For a circular orbit, the aerothermal flux decrease with the altitude
from 1300W/m
2
at 150km to 160 then 39W/m
2
at 200 and 250 km. The
remaining aerothermal flux at 350 km is only 7W/m
2
. This contribution
will then be neglected throughout this work.
3.3.6 Results
Using this model, we are now able to represent the fluxes seen by the
satellite on its orbit. In Figure 3.5 is plotted the nominal Earth IR and
albedo fluxes incident to a unit surface pointing toward the Earth (ρ =
180

), on OUFTI-1’s orbit, Ω = 0

(line of nodes perpendicular to the sun
vector), ω = 0

.
0 10 20 30 40 50 60 70 80 90 100
0
50
100
150
200
250
300
350
400
Time [min]
I
n
c
i
d
e
n
t

f
l
u
x
e
s

[
W
/
m
²
]


Albedo
Earth infrared
Figure 3.5 – Nominal Earth IR and Albedo uxes seen by a surface pointing to the
Earth on OUFTI-1’s orbit
But it has already been noticed that when the satellite is close to the
earth, unless a face points perfectly away from the Earth, all the six faces
have a non zero view factor F
E
. This means that an effective area can be
defined as the area pointing toward the Earth that would receive the same
flux as the sum of all the faces.
A
e f f
=
1
q
nom
n
f aces

i
A
i
q
i
=
A
q
nom
n
f aces

i
q
i

A
e f f
A
=
1
q
nom
n
f aces

i
q
i
The figure 3.6 shows this. The CubeSat is initially oriented in such a
way that its great diagonal is parallel to the local zenith. For this position,
all the six faces have a non-zero incident flux and that the effective area
ratio (ratio between the effective area and a cube’s face) is greater than 2 !
Lionel Jacques
2
nd
Master in Aerospace Engineering
27 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
0 10 20 30 40 50 60 70 80 90 100
0
50
100
150
200
Time [min]
I
n
c
i
d
e
n
t

i
n
f
r
a
r
e
d

f
l
u
x

[
W
/
m
²
]


Face 1
Face 2
Face 3
Face 4
Face 5
Face 6
0 10 20 30 40 50 60 70 80 90 100
0
0.5
1
1.5
2
2.5
Time [min]
E
f
f
e
c
t
i
v
e

a
r
e
a

r
a
t
i
o
Figure 3.6 – Infrared uxes incident on each face and effective area ratio
The case of a randomly spinning CubeSat is represented in Figure 3.7.
The effective area ratio does not significantly change and this is the same
for albedo and IR. The conclusion is that the spin rate of the CubeSat does
not significantly affect the effective area ratio related to IR and Albedo for
LEO.
0 10 20 30 40 50 60 70 80 90 100
0
0.5
1
1.5
2
2.5
Time [min]
E
f
f
e
c
t
i
v
e

a
r
e
a

r
a
t
i
o


Solar
IR
Albedo
Figure 3.7 – Effective area ratio
On another hand, the effective area ratio related the solar flux never
exceeds the theoretical value of

3, corresponding to the case when the
Lionel Jacques
2
nd
Master in Aerospace Engineering
28 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
great diagonal is pointing to the sun (the projected area is

3 times greater
than the cube’s face area).
3.4 Cyclic Transient Model
Now that the incident fluxes on each face can be computed, the tem-
perature must be computed. For this preliminary analysis, a very simple
model has been created. It is based on only one node for all the satellite
(so considered isothermal) but taking transient effects into account.
3.4.1 Implementation
The basic underlying equation used for this model is the lumped pa-
rameter equation, written here under its general form for a transient mul-
tiple nodes model :
Qint
i
+ Qext
i
+
n

j=1,j =i
GL
i,j
(T
i
T
j
) + σ
n

j=1,j =i
GR
i,j
(T
4
i
T
4
j
) = C
i
dT
i
dt
(3.3)
Qint
i
the total internal dissipated power for the node i, Qext
i
the to-
tal external power incoming on the node i (Qext
i
= QS
i
+ QA
i
+ QE
i
),
GL
i,j
and GR
i,j
are respectively the conductive and radiative links between
nodes i and j. T
i
is the temperature of the node i and C
i
its heat capacity.
This equation results from the heat balance to node i.
For a single node model, it reduces to :
Qint + Qext + σGR
sat,DS
(T
4
sat
T
4
DS
) = C
dT
sat
dt
where GR
sat,DS
is the radiative exchange factor between the satellite
and the Deep Space and T
DS
= 3K the Deep Space temperature. As there
is no coupling between the faces of the cube, GR
sat,DS
is simply equal to
the product of the CubeSat’s total area by its averaged emissivity :

eq
=

i
A
i

i

i
A
i
Concerning Qext, the heat flux incident on face i are simply multiplied
by its corresponding equivalent absorptivity/emissivity (α for solar and
albedo and for IR) and then added:
Qext = QS + QA + QE =
n
f aces

i
A
i

i
q
S,i
+ α
i
q
A,i
+
i
q
E,i
)
The cyclic transient solver is implemented in Matlab/Simulink envi-
ronment. The solver is made up two main consecutive phases : a first
one calculating the external incident fluxes around the orbit in Simulink
(already developed and represented at the figure 3.2) and a second one
integrating the lumped parameter equation. Once the first phase is com-
pleted and the external absorbed power is known for each face and at
each position around the orbit, the equation 3.3 is integrated over one or-
bit several times. By this way, after a finite number of orbit integrations,
Lionel Jacques
2
nd
Master in Aerospace Engineering
29 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
the temperature evolution over one orbit does not depend anymore on the
first initial condition. The process is summed up in the flow chart of the
figure 3.8.








Equations of motion
integration with Simulink
during one orbit
Solar, Albedo and IR incident
fluxes computation
Total effective absorbed
power on one orbit
Thermal transient
equation integration
during one orbit
Set initial
0

Set
0
+1
=

Converge ?
No
Yes
()
Figure 3.8 – Cyclic transient Matlab Simulink solver ow chart
The cyclic repetition of the temperature evolution is based on two con-
vergence criteria.
1. The temperature reached at the end of one orbit must be equal to
the temperature at the beginning of this orbit. If they are not equal,
the initial condition of the (k +1)
th
orbit integration is updated with
the final temperature of the previous orbit.
_
_
_T
k
t=0
T
k
t=T
orbit
_
_
_ < tol
T
where T
orbit
is the period of the orbit.
2. The time derivative of the temperature at the beginning and at the
end of one orbit must be equal.
_
_
_
_
_
dT
k
dt
¸
¸
¸
¸
t=0

dT
k
dt
¸
¸
¸
¸
t=T
orbit
_
_
_
_
_
< tol
dT
In practice, the Runge-Kutta algorithm (ode45 in Matlab) is used to
solve the non linear differential equation over the orbit period. If the con-
vergence criteria are not satisfied, the initial temperature of the next orbit
is set as the final one of the previous orbit, as described in the flow chart.
3.4.2 Validation
Before running and analyzing the model, it has been correlated with
ESATAN/ESARAD softwares. The validation is performed for a non-
rotating cube pointing toward the sun on OUFTI-1’s orbit (354 1447km
Lionel Jacques
2
nd
Master in Aerospace Engineering
30 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
71

), with a RAAN Ω = 45

, and a perigee argument ω = 0

. The cube is
considered as a perfect black-body (α = = 1). To evaluate its heat capac-
ity, the heat capacities of OUFTI-1’s main components
1
have been added.
The different contribution are exposed in the table 3.2.
Component Mass [g] Specific Heat [J/kgK] Heat capacity [J/K]
aluminum frame 145 980 142.1
aluminum panels 165 980 161.7
PCBs 350 1136 397.6
Antenna panel 33 980 32.3
Batteries 50 960 42.3
Total 698 776
Table 3.2 – OUFTI-1s thermal heat capacity
The figure 3.9, shows that both temperature and flux are identical with
the two solvers and that the Matlab/Simulink one is valid.
0 10 20 30 40 50 60 70 80 90 100
−30
−25
−20
−15
−10
−5
0
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


Matlab/Simulink
ESATAN/ESARAD
0 10 20 30 40 50 60 70 80 90 100
0
5
10
15
Time [min]
P
o
w
e
r

[
W
]


Matlab/Simulink Sun
ESATAN/ESARAD Sun
Matlab/Simulink Albedo
ESATAN/ESARAD Albedo
Matlab/Simulink Infrared
ESATAN/ESARAD Infrared
Figure 3.9 – Validation of the cyclic transient Matlab/Simulink solver with
ESATAN/ESARAD
3.5 Results
Now that one can convincingly compute the mean temperature of the
CubeSat for given thermo-optical properties and orbit, it is time to briefly
1
All materials properties data comes from CES EduPack. CES EduPack, created by
Professor Mike Ashby of Cambridge University, is a very useful tool for materials selection,
containing an impressive database of materials and process properties.
Lionel Jacques
2
nd
Master in Aerospace Engineering
31 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
introduce the concept of cold and hot case, as it will be reused many times
throughout this work. It is based on a worst case conservative study :
the hot case corresponds the case for which all parameters, both
environmental and satellite related, are chosen in such a way that
they all contribute to reach maximal temperatures and/or gradients
during one orbit. For instance, considering the solar constant at the
winter solstice (1414 [W/m
2
]) instead of the mean value 1367 [W/m
2
]
or a higher albedo coefficient or even a particular orbit (as the orbit
changes over time).
On the contrary, the cold case is the perfect opposite of the hot case:
all parameters are such as the reached temperature is minimal.
The parameters that are biased hot or cold are:
Emissivity and Absorptivity
Environmental Constants (Solar, Albedo, IR)
Spacecraft Orbital Orientation
Internal Power Dissipation
External surfaces emissivity/absorptivity values are not perfectly
known and vary throughout the lifetime of the spacecraft (mainly an in-
crease for α due to high reactive atomic oxygen present in the upper layers
of the atmosphere) and are therefore subjected to the cold/hot case defini-
tion. The concept of Beginning of Life (BOL) and End of Life (EOL) values
is then useful. As the emissivity does not suffer great variation over time
while the absorptivity tends to increase, BOL values are considered for the
cold case analysis and EOL for the hot case. But at this point, as these val-
ues are not yet perfectly knows and defined, this effect will not be taken
into account. A sensitivity analysis will be performed to see the relative
importance of each thermo-optical property.
The cold and hot case assumptions are summarized in the following
table 3.3:
Parameter Hot Case Cold Case
Orbital parameters permanently illuminated max eclipse time
Solar constant 1414 [W/m
2
] 1322 [W/m
2
]
Albedo coefficient 0.35 0.25
Earth temperature 250K ( 220[W/m
2
]) 260K ( 260[W/m
2
])
Internal dissipation full none
Table 3.3 – Preliminary Hot & Cold case definition
As described in the table, for the cold case, all the electrical power col-
lected by the solar cells is assumed to be perfectly stocked in the batteries
without any dissipation. On the other hand, it is totally converted into
heat in the hot case (cells’ efficiency equal to zero). These are indeed the
worst cases. One finally considers that the satellite is randomly rotating
for both cases
1
.
1
ADCS simulations [25] shows that the satellite will never stop rotating because of the
passive control and continuous perturbation torques acting on it
Lionel Jacques
2
nd
Master in Aerospace Engineering
32 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
3.5.1 Hot Case
As explained above, the first hypothesis used for the hot case is a con-
stantly illuminated orbit which is possible due to the high inclination of
OUFTI-1’s orbit : 71

. This means that the line of nodes, defined by the
RAAN, is perpendicular to the direction of the sun. But in this situation,
albedo is reduced because the satellite is continuously far from the sub-
solar point. One could wonder if an orbit containing the subsolar point
with a minimum eclipse time (perigee in eclipse) could be more critical.
The figure 3.10 shows the influence of the orbit orientation on the hot case
temperatures evolution.
Figure 3.10 – Hot Case definition, in uence of orbit orientation
One can see that the lower the angle between the line of nodes and the
sun direction, the lower the temperature reached. But there’s an angle near
70

for which the satellite still experiences no eclipse and yet have a slightly
higher albedo. This case would be slightly worst than the presupposed
one but the difference is small and the hot case will still consider an orbit
whose line of nodes is perpendicular to the sun’s direction.
Recalling the maximal specified temperature of the batteries of 45

C,
the mean temperature of the CubeSat in this preliminary hot case is al-
ready close to it and will thus be carefully looked at in the next analysis.
3.5.2 Cold Case
Here are the results of the cold case defined earlier. The temperature
falls down to -20

C. This would represent the mean temperature of the
CubeSat and one can imagine that external parts could still be cooler and
internal parts slightly hotter. When the satellite comes out of eclipse, the
batteries begins to charge and this analysis let us think that the batteries
could be below their minimal specified temperature while charging: 0

C.
Lionel Jacques
2
nd
Master in Aerospace Engineering
33 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
0 10 20 30 40 50 60 70 80 90 100
−20
−15
−10
−5
0
5
10
Time [deg]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]
Figure 3.11 – Cold Case temperature evolution
3.6 Sensitivity analysis
Except for the solar cells, absorptivity/emissivity values are not yet
well known. That’s why, we carried out an evaluation of the sensitivity of
the extremal temperatures to thermo-optical properties. One can already
compute the area ratios of the different external materials of the CubeSat.
The results are given in the table 3.4.
Component Area [cm
2
] Ratio [%]
Solar cells 301.8 50.3
aluminum panels 107.2 17.9
Anodized rails 68 11.3
Remaining aluminum frame 65.6 10.9
Antenna deployment mechanism panel 57.4 9.6
Total area 600 100
Table 3.4 – External area ratios
The solar cells obviously fill the half of the total area of the CubeSat
while the aluminum panels and rails cover 30% of the cube. The remaining
20% are divided up between the frame and antennas’ panel.
To do so, we vary the α/ values for each component separately and
compute the minimal/maximal temperatures in the cold/hot case, T
min
&
T
max
. For instance, the figure 3.12 represents the evolution of T
min
& T
max
for all combination α/ of the aluminum panels.
Figure 3.12 shows that all the curves are nearly linear. Linearized sen-
sitivities of the maximal and minimal temperatures of the CubeSat to the
emissivity and absorptivity of the aluminum panel can thus be computed.
The coefficients for the other surfaces are obtained by the same way: all
Lionel Jacques
2
nd
Master in Aerospace Engineering
34 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
0 0.2 0.4 0.6 0.8 1
15
20
25
30
35
40
45
50
T
m
a
x

[
°
C
]
Aluminium panel emissivity
0 0.2 0.4 0.6 0.8 1
−35
−30
−25
−20
−15
−10
T
m
in

[
°
C
]
Aluminium panel emissivity


α=0.1
α=0.2
α=0.3
α=0.4
α=0.5
α=0.6
α=0.7
α=0.8
α=0.9
Figure 3.12 – T
max
& T
min
sensitivity to aluminum panel thermo-optical properties
the values are exposed in table 3.5. The value must be interpreted as the
variation of the minimum/maximum temperature resulting from a 10%
variation of the thermo-optical property. For instance, if the absorptivity
of the rails increase(decrease) from 10%, the maximal temperature will
decrease(increase) from 0.8

C.
dT
max
d
0.1
dT
min
d
0.1
dT
max

0.1
dT
min

0.1
aluminum panels -1.8 -1.4 1.8 1.4
Rails -0.8 -0.6 1.0 0.8
aluminum frame -1.0 -0.8 0.6 0.4
Antennas’ panel -0.9 -0.8 0.1 0.1
Table 3.5 – Sensitivity coefficients to thermo optical properties
It emerges that the aluminum panels properties have the more influ-
ence but that it yet remains relatively small.
3.7 Advanced Simulink Model
As presented in the introduction, V. Beukelaers, who is in charged of
the MIAS subsystem, develops a global simulator in Matlab Simulink envi-
ronment. Within this framework, a more detailed simulink thermal model
has been developed. However, for the sake of conciseness, only its main
features and results are presented. Each PCB and face has a dedicated
node in addition to the batteries’ node. The incident fluxes are computed
though the same previously used module.
Figures 3.13 and 3.14 display the results for the hot and cold cases. The
bottom plots show the evolution of the eclipse time while the two others
Lionel Jacques
2
nd
Master in Aerospace Engineering
35 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
represent the evolution of the temperature of the PCBs and faces. Figure
3.13 shows the reduction of eclipse time and the resulting transient period
lasting four to three orbits before the hot case is achieved.
Figure 3.13 – Advanced simulink thermal model hot case results
Figure 3.14 – Advanced simulink thermal model cold case results
Lionel Jacques
2
nd
Master in Aerospace Engineering
36 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 3. Preliminary thermal analysis
Summary
The preliminary analysis already suggests that the battery, which has
the narrower allowable temperature range, will be one of the key driver of
the thermal design in the hot case as well as in the cold case. On another
hand, hot spots could occur in the hot case but are not observable here.
Moreover, the available surface area for thermal control coating is low
and their individual effect on the extreme reached temperatures are quite
small. However, accurate data about the optical-properties of the outer
surfaces should be gathered
Lionel Jacques
2
nd
Master in Aerospace Engineering
37 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
4
Simplified Thermal Model
After this first preliminary analysis, the use of ESATAN/ESARAD soft-
wares seemed necessary even for the development of a simple multi-nodes
model. This chapter includes a brief description of these softwares and
thermal modeling. The particular modeling flow chart will be described.
Then, it will be used for the creation of our first multi-nodes model,
namely the Simplified Thermal Model (STM). The aim of this model is
not to give accurate temperature distributions of PCBs but rather a first
guess of the temperature of OUFTI-1’s main units which was not available
in the previous single-node model.
4.1 Thermal modeling with ESATAN & ESARAD
The global thermal analysis is performed in two stages : the radiative
analysis and the thermal analysis. These two stages are performed respec-
tively with ESARAD and ESATAN softwares, ESA’s standard tools for
thermal radiation and thermal analysis. It is based upon the lumped pa-
rameter method. The fundamental assumption is that it considers isother-
mal nodes in a thermal network (electrical analogy).
The global thermal model is thus divided up into two separated
models with different purposes : the Geometric Mathematical Model
(GMM) and the Thermal Mathematical Model (TMM). GMM’s goal, cre-
ated within ESARAD, is to compute the radiative coupling between its dif-
ferent nodes and the environmental thermal fluxes along the orbit which
are then input into the TMM. The TMM includes conductive links and
heat capacity and is solved within ESATAN to predict the temperatures of
the spacecraft.
For our model creation, pre/post-processing is done in Matlab envi-
ronment. This choice is motivated by relatively low parametrization of
ESATAN/ESARAD models. GMM and TMM input files are written within
Matlab, based on spacecraft and environmental data, all parametrized.
ESARAD/ESATAN are then launched within Matlab, using batch mode,
except for model checking. All output files are also post-processed within
Matlab. By this way, parametric studies are easier to perform since all the
process is Matlab controlled. Figure 4.1 shows the thermal model process-
ing flow chart.
The conductive network links and nodal heat capacities are calculated
by hand, based on material properties and geometrical configuration.
38
Chapter 4. Simplified Thermal Model

GR & Fluxes
GMM
TMM
GL, QI & C
ESARAD
ESATAN
Temperatures
Post processing
Geometry,
Environment
Matlab
Matlab
Matlab
Figure 4.1 – Thermal model processing ow chart
Lionel Jacques
2
nd
Master in Aerospace Engineering
39 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
4.2 Geometric Mathematical Model
4.2.1 Nodal breakdown
For a good understanding and easy interpretation of heat flow paths,
this first multiple-nodes model remains relatively simple. It is made up of
18 nodes :
One node per face : each face is considered as one node and so
isothermal, for all face’s layers except solar cells. We call layer the dif-
ferent elements constituting the face along its thickness: successively
the aluminum frame, the aluminum panel and the solar cells. This
is a relatively good assumption seeing the small thickness and high
conductivity of aluminum. The glue is neglected.
One node for each pair of solar cells (two per face).
One node per PCB (OBC, OBC2, EPS, Batteries’ PCB, EPS2, COM),
representing the average temperature of the PCB.
One node for the two batteries.
Here is the numbering convention adopted :
- i
th
face solar cells 10i i = 1 . . . 6, i = 4
- i
th
Face 30i i = 1 . . . 6
- OBC 10000
- OBC2 12000
- EPS 14000
- BAT 15000
- EPS2 16000
- COM 18000
- Batteries 20000
Figure 4.2 illustrates the nodal breakdown and GMM. Spacers and PCB
connectors have not been modelled, considered just as conductive links.
Neglecting spacers is a good assumption seeing their low contribution
in the radiative model (low area and low emissivity of aluminum) and
low capacitance for the TMM. The assumption about the connectors is
stronger : in the GMM, the view factors between face 5 and PCBs are
artificially increased but as it will be shown, radiation inside the CubeSat
is less influent than conduction. Nevertheless, their heat capacity will be
taken into account and included to the PCBs.
4.2.2 Thermo-optical properties
Concerning thermo-optical properties, precise measurements are not
affordable for a student project. Therefore, data have been collected from
many sources and compared in order to choose the most realistic values
as possible. Recall that the CubeSat’s rails are hard anodized to prevent
Lionel Jacques
2
nd
Master in Aerospace Engineering
40 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
Figure 4.2 – Geometric Mathematical Model nodal breakdown in ESARAD
galling while the rest of the frame is alodyned to enhance electrical con-
ductivity
1
. Both processes will be studied in details in the next chapter.
aluminum panels’ surface treatment is also Alodine (1200). For insulation
reasons, solar cells have not been directly integrated on the panels but a
Kapton
R
foil is used as insulator and completely covers the panels. Data
for the kapton
TM
covered panels comes directly from EADS Astrium, who
performed the solar cells integration on the aluminum panels, and did op-
tical properties measurements of the integrated panels. Figure 4.3 shows
the panels before and after solar cells integration.
Figure 4.3 – On the left, the CubeSat with one aluminum panel before integration. On
the right, the panel integrated with the cells by EADS Astrium
4.2.3 Orbit & attitude
As already said in the beginning of this work, OUFTI-1 is selected for
the Vega Maiden Flight and the only defined orbital parameters are :
1
According to Adam Reif from Pumpkin, the MIL-spec reference of the anodized pro-
cess is MIL-A-8625F, TYPE III, CLASS 1 while the one of alodyned process is MIL-DTL-
5541F Type II, CLASS 1A
Lionel Jacques
2
nd
Master in Aerospace Engineering
41 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
Part Material Thermal finish α reference(s)
aluminum
frame
5052 aluminum
alloy
Alodine
0.08 0.15 [27]
0.08 0.15 [11]
0.1 0.1 [57]
0.08 0.15 [20]
aluminum
frame rails
5052 aluminum
alloy
Hard anodized
0.88 0.88 [24]
0.86 0.86 [50]
0.85 0.87 [49]
0.86 0.86 [43]
aluminum
panels, inside
7075 aluminum
alloy
Alodine 1200
- 0.1 [45]
- 0.08 [8]
- 0.07 [43]
aluminum
panels, out-
side
7075 aluminum
alloy
2mil Kapton
R
foil
on Alodine 1200
0.87 0.81 (a)
Solar cells triple junction
GaAs cells
anti reflective coat-
ing
0.91 0.81 (a), [7]
PCBs FR4 - - 0.8 [64]
Battery pack plastic - - 0.8 [21]
Notes : (a) : EADS Astrium measurements
Table 4.1 – Thermo Optical Properties
its apogee: 354 km,
its perigee: 1447 km
and its inclination: 71

.
Due to the telecommunication payload, the satellite does not require
a precise attitude control so that this will be a passive one. As described
in the ADCS subsystem presentation, a permanent magnet will be used
to align the CubeSat on Earth’s magnetic field and hysteretic materials to
damp the remaining rotation energy.
In ESARAD, this attitude has been simplified by using a planet cen-
tered Local Orbit Coordinate System (LOCS), described on the figure 4.4.
Within this coordinate system, as the orbit inclination is relatively high,
the alignment on Earth’s magnetic field is approximated by prescribing a
constant rotation speed around y axis such as the CubeSat perform two
revolutions during one orbit. The LOCS already performing one revolu-
tion along the orbit, the prescribed rotation rate is
360
T
deg/sec where T is
the period of the orbit.
4.2.4 Radiative coupling
The calculation of view factors and direct heat fluxes is performed in
ESARAD by the Monte Carlo ray-tracing (MCRT), also used in realistic
rendering. MCRT method is based on a statistical approach. For a perfect
Lionel Jacques
2
nd
Master in Aerospace Engineering
42 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
page 6-13

Figure 6-8 LOCS orientation options
6.2.2.1 Planet-centred orbits
A planet-centred orbit may be planet-oriented, True-Sun-oriented or Sun-oriented.
PLANET_ORIENTED +Z points towards the local zenith, +X is in the orbital plane,
perpendicular to +Z and in the direction of the velocity, and
Figure 6-9 Planet-oriented LOCS
Orbits may be:
Planet-centred
Sun-centred
In ESARAD language:
TRANSFER_ORIENTED
Planet-oriented
True-Sun-oriented
PLANET_ORIENTED
TRUE_SUN_ORIENTED
Sun-oriented SUN_ORIENTED
North
Reference line
Z
X
Y
Z
X
Y
Reference plane
Figure 4.4 – Planet centered Local Orbit Coordinate System (LOCS), from ESARAD
User Manual
accuracy, an infinite number of rays should be fired from each face since
each ray has its own path, its own initial direction and may reach and be
reflected by any other radiative face in the model. Firing and following the
path of an infinity of rays is not possible. The statistical feature of MCRT
comes from the estimation of the radiative couplings or heat fluxes made
by averaging the results obtained from a finite random sample of rays.
page 7-5

of rays, an estimate of the radiative couplings or heat fluxes can be made by averaging the
results obtained from a finite random sample of rays. This is the basis of the Monte Carlo
method. The individual history of each ray, that is its emission point, emission direction and
ray/face interaction, is randomly determined. MCRT is, therefore, essentially a stochastic
method.
MCRT is a very powerful simulation method. It allows the modelling of the thermal
radiative behaviour of the system without introducing any special assumptions or
restrictions, other than those derived from the idealisation used to describe the ray emission
and the ray/face interaction. Diffuse emission and reflection, specular reflection, and
transmission are thus modelled exactly.
The high fidelity of the method is, however, obtained at a cost. Accurate MCRT runs tend to
be time consuming, especially when compared with the alternative matrix methods. This is
especially true for models with low-emissivity faces, since the number of multireflections is
dramatically increased.
Another characteristic of MCRT runs is their randomness. Concerning the accuracy of the
results, the only definitive statement that can be made is that the evolution of the estimated
value is within a band whose width is inversely proportional to the square root of the
number of rays fired. MCRT being a random process, it is absolutely impossible to predict
the evolution of the estimated value within the band (see Figure 7-2).
7.4.1.1 MCRT pre-processing
When you perform an MCRT calculation, ESARAD pre-processes the geometric model to
convert it into a format which is more suitable for ray tracing. In a nutshell, the space
occupied by the model is broken down into volume elements, and the model’s radiative
faces are classified with respect to these volumetric cells (known as voxels).
Figure 7-2 Evolution of MCRT-calculated entities
e
k
N
-------- =
seed
2
seed
1
theoretical
value
Figure 4.5 – MCRT convergence from ESARAD user manual [2]
The two control parameters of MCRT method are the number of rays
fired from each face and the seed, a user input defining the pseudo-random
initial distribution of rays’ orientation. Due to the stochastic nature of
MCRT, different seed values will lead to different results, as shown by
Figure 4.5. However, the estimated value evolution goes randomly within
a band whose width is inversely proportional to the square root of the
number of rays fired. A disadvantage of MCRT method is that required
calculation time increase as the face emissivity/absorptivity decrease since
the rays are reflected more times before being absorbed.
This random evolution could lead to some mistakes and a convergence
analysis was performed. Concerning our GMM, external faces have no ra-
diative coupling so that environmental fluxes calculation does not require
many rays. On another hand, figure 4.6 illustrates the convergence of the
radiative exchange factor (REF) between the EPS PCB and face 1 for dif-
ferent seeds. By setting the number of rays to 10000 (or more), figure 4.6
shows that the relative error is about 1% (or below).
Lionel Jacques
2
nd
Master in Aerospace Engineering
43 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
10
3
10
4
10
5
10
6
0
0.5
1
1.5
2
2.5
3
e
r
r
o
r

o
n

R
E
F

b
e
t
w
e
e
n

f
a
c
e

1

a
n
d

E
P
S

P
C
B

[
%
]
Number of rays fired, log scale


seed 1
seed 2
seed 3
Figure 4.6 – Radiative exchange factor convergence with number of rays fired (MCRT)
4.3 Thermal Mathematical Model
Now that the GMM is defined and ready for computation, the second
stage of the thermal analysis consist in the TMM creation. TMM involves
conductive links network and nodal heat capacities definition and internal
heat load distribution. These tasks are made by hand within Matlab and
described for each group of nodes in the next paragraphs.
4.3.1 The structure
As previously described, each face of the CubeSat is modelled by one
node. This means that temperature gradients inside a face are assumed
to be low and therefore neglected : the face is considered isothermal. The
conductive flow path between the faces only goes through the aluminum
frame since there is no other contact. This is an approximation : the con-
tribution of aluminum panels to the conductance between faces has been
neglected. Seeing the relatively complex geometry of the frame, conduc-
tive links calculation is not immediate. Hence, for each of the 12 links, a
finite element model has been created within SamcefField.
The method used is the following one : to compute the equivalent co-
ductive link between two faces, each of them is cut in two in the middle,
perpendicular to the flow path. A temperature gradient is imposed across
the two half faces and SamcefField is then able to compute the resulting
flux proportional to the temperature gradient. The equivalent conductance
is easily deduced from GL =
Q
∆T
. Figure 4.7 shows the temperature distri-
bution across the face 1 and 3 for a gradient of 10

C.
The 12 conductances are exposed in the table 4.2.
These values do not take into account the contact resistance in the
links with faces 3 and 6. Contact results from the fact that faces 3 and 6
Lionel Jacques
2
nd
Master in Aerospace Engineering
44 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
Figure 4.7 – Evaluation of the conductive link of aluminum frame through a finite
element analysis within SamcefField
Faces GL [W/K] faces GL [W/K]
GL
12
5.496 10
2
GL
26
5.465 10
2
GL
13
5.465 10
2
GL
34
5.789 10
2
GL
15
5.496 10
2
GL
35
5.465 10
2
GL
16
5.465 10
2
GL
45
5.924 10
2
GL
23
5.465 10
2
GL
46
8.988 10
2
GL
24
6.519 10
2
GL
56
5.465 10
2
Table 4.2 – Equivalent conductance between aluminum frame faces
are fixed on the lateral frame with screws : one for each face, except for
face 4 fixed with 3 screws because of the ports. The two assumptions tends
to balance each other since that neglecting contact overestimate the con-
ductance while neglecting panels conduction contribution underestimate
the total conductance. These phenomena will be studied and taken into
account in a more detailed model : the low level model.
The frame is made of 5052 aluminum alloy while 7075 aluminum alloy
was used for aluminum panels. Properties of these alloys are given in the
table 4.3 and comes from the CES EduPack software.
Alloy ρ [kg/m
3
] c [J/kg.K] k [W/m.K]
5052 2672 - 2698 963 - 1002 140 - 152
7075 2770 - 2830 913 - 979 131 - 137
Table 4.3 – aluminum alloys properties
Using these properties, the nodal capacitances of the faces are com-
puted by adding the capacitances of the frame face and associated panel.
Lionel Jacques
2
nd
Master in Aerospace Engineering
45 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
Node C [J/K]
301 46.5
302 46.5
303 58.9
304 48.1
305 46.5
306 63.9
Table 4.4 – Nodal capacitances
4.3.2 The solar cells
Electrical power management
As the SwissCube, a dissipation circuit has been designed by P. Ledent
and P. Thirion in order to deal with the extra power produced by solar
cells when the batteries are charged and the required power is lower than
collected one. The only way of dissipation in space is heat. As the Cube-
Sat’s attitude is passively controlled, no face is pointing to deep space and
the use of a radiator to ensure heat rejection is not to be foreseen. In a first
design, the extra power is thus dissipated through electrical resistances on
the EPS PCB.
The energy collected by the cells is not directly available in ESARAD
but can be easily obtained from the power balance of a solar cell (figure
4.8).
d
d
d
d
d

Q
in
= AC
s

Q
re f l
= (1 α)Q
in
d
d
d
d
d

Q
Elec
= η(T)Q
in

©
Q
Ther
Figure 4.8 – Solar cell power balance
Q
in
is the incident power, equal to the projected area A multiplied
by the solar constant C
s
and η is the solar cell efficiency. The efficiency
available in the data sheets is indeed the AM0 efficiency, meaning that it
refers to the incident power and not the absorbed one. The equilibrium
gives:
Q
in
= (1 α)Q
in
+ Q
Ther
+ Q
Elec
Lionel Jacques
2
nd
Master in Aerospace Engineering
46 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
Therefore, the heat generated by the cells is given by:
Q
Ther
= (α η(T))Q
in
But ESARAD gives us the absorbed power : Q
ESARAD
= αQ
in
and here
is the electrical collected power Q
Elec
expressed in function of Q
ESARAD
:
Q
Elec
= η(T)Q
in
=
η(T)
α
Q
ESARAD
In ESATAN, this power is subtracted at the cells and reinjected, par-
tially or not, inside the CubeSat, where dissipation occurs.
Solar cells' model
Solar cells have a dedicated node with its own capacitance and its link
with the corresponding face. The layout of the solar cell integration made
by EADS Astrium is the following one : aluminum panels have been cov-
ered with 50 microns (2mil) thick Kapton
R
foil for insulation reason. The
laydown adhesive is RTV S 691 (silicone adhesive) for the Kapton
R
foil
and for the solar cells and is 80 microns thick. From this layout, the equiv-
alent thermal resistance between the cell and the face becomes :
1
GL
cell, f ace
= R
cell, f ace
=
1
A
cell
_
t
cell
k
cell
+
t
adhesive
k
adhesive
+
t
Kapton
k
Kapton
+
t
adhesive
k
adhesive
+
t
aluminum
k
aluminum
_
Kapton
R
and RTV S 691 properties are given in the table 4.5
k [W/m.K]
Kapton
R
0.12
RTV S 691 0.39
Table 4.5 – Thermal conductivity of Kapton
TM
[14] and RTV S691 adhesive [18]
The cells are 150 microns thick [7], have a thermal conductivity of more
or less 100 [W/mK] [59], a specific heat of about 700 [J/KgK] [59] and
weights 86 [mg/cm
2
][7].
4.3.3 The PCBs stack
As explained previously, each PCB is modelled with only one node,
representing the mean temperature of the PCB and this model will not
be able to reveal local hot spots inside the PCBs. The heat capacity of the
PCBs is evaluate by multiplying their mass (about 70g, including PCB and
connector) with a equivalent specific heat. The equivalent specific heat is
determined with the help of the online calculator developed by Frigus
Primore [46].
Lionel Jacques
2
nd
Master in Aerospace Engineering
47 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
Inter PCBs links
Concerning the conductive links between the PCBs, there are two main
heat flow paths : through the spacers and through the connector. Accord-
ing to the PC/104 specification [41], the connector is made up of 104 phos-
phor bronze pins and only pins contributes to the link since connector’s
housing height is such as there is no contact with the above PCB.
Conduction through the M3 endless screws is neglected seeing the
clearance with the PCB holes and spacers. The spacers are made of 6061-
T6 aluminum and the contact between spacers and PCBs is considered
to be perfect. A spacer is a small hollow aluminum cylinder with diam-
eters of 3.3mm (interior) and 4.5mm (exterior). Thermal conductivities of
phosphor bronze and 6061-T6 aluminum are given in table 4.6 :
k [W/m.K] reference
Phosphor bronze 75 ECSS Q-70-71A[18]
6061-T6 aluminum 152 - 169 CES EduPack
Table 4.6 – PCBs links materials thermal conductivities
The ratio between the link through spacers and the link through con-
nector is:
4 S
spacers
k
spacers
104 A
1pins
k
pins
=
4 7.35 10
6
160.5
104 0.508 10
6
75
= 1.2
This means that conduction through the spacers is as well important
as conduction through the connectors.
PCBs stack links with the structure
The PCBs stack has 8 contact point with the external structure : four at the
top with the COM PCB and 4 at the bottom with OBC PCB. The top links
are ensured bu the "midplane standoffs" as presented on the exploded view
in chapter 2. Two midplanes ensure the fixation with the face 5 and two
with face 2. The conductive link through is not easy to evaluate and, as for
the aluminum frame links, a finite element model has been created within
SamcefField. Again, perfect contact is assumed between the PBC and the
frame.
Figure 4.9 illustrates the temperature distribution inside a midplane
standoff when a gradient of 10

C is imposed between the two perpendic-
ular faces involved. The equivalent thermal conductance is obtained by
dividing the resulting flux by the difference of temperatures.
Concerning the bottom link of the PCBs stack, OBC PCB is supported
by four stainless steel threaded loose fasteners (in which the endless
screws are screwed) inserted in the bottom frame face. Once again, perfect
contact is assumed.
4.3.4 The batteries
The two LiPo batteries are located on a secondary smaller PCB itself
fixed on the EPS PCB as shown by the exploded view (figure 1.11). The
Lionel Jacques
2
nd
Master in Aerospace Engineering
48 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
Figure 4.9 – Conductive link computation through a finite element analysis within
SamcefField of a midplane standoff
GL [W/K]
Structure - OBC 0.12
OBC - OBC2 0.58
OBC2 - EPS 0.58
EPS - BAT PCB 0.35
EPS - EPS2 0.35
EPS2 - COM 0.58
COM - face 2 1.42
COM - face 5 1.42
Table 4.7 – PCBs links
conductive link between the batteries’ PCB and the EPS PCB is also en-
sured by four spacers. The material of these spacers is not yet defined and
is kept as parameter. aluminum conductivity will be assumed as initial
value.
Concerning the specific heat of the batteries, we used the one of Varta
PoLiFlex
R
available in the SwissCube Phase C technical report [35], keep-
ing in mind that the model used is not yet defined.
4.4 Worst cases definition
As introduced in the previous chapter, thermal analysis is subjected to
a worst cases approach.
The assumptions about cold and hot cases are the same as the one
described in the preliminary analysis :
- Hot case : maximum environmental fluxes, orbit permanently illumi-
nated, electrical collected power entirely dissipated into heat. Now
that PCB are modelled, the internal power is distributed between
the two PCBs containing the most dissipative components : COM
(amplifier) and EPS (dissipation circuit, voltage converter). Without
Lionel Jacques
2
nd
Master in Aerospace Engineering
49 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
accurate information about these components, an equal distribution
between COM and EPS has been assumed.
- Cold case : minimum environmental fluxes, maximum eclipse time
(apogee in eclipse involving more than 35 min in eclipse for a period
of 104 min), no internal dissipation.
The effect of thermo-optical properties degradation is only taken into
account for the Kapton
R
used on the panels, its absorptivity increasing
over time, and the solar cells efficiency, decreasing over time, since alu-
minum surface treatments are relatively stable over a lifetime such as
the one expected for OUFTI-1 (about one year). As the absorptivity of
Kapton
R
, measured by EADS Astrium, is already high (0.87) the End-Of-
Life value is set to 0.9. Concerning the solar cells efficiency, a diminution
of 10% is considered : 30% BOL and 27% EOL.
4.5 Results
4.5.1 Hot case
Figure 4.10 shows the different contributions to the total absorbed
power during one orbit. The total power is represented by the black dotted
line and the effective thermal power, obtained by subtracting the electrical
power collected by the cells from the total power, is the continuous black
line. As explained above, all the electrical power is redistributed inside the
CubeSat as internal heat loads so that the total power (dotted line) finally
contributes to the heating of the CubeSat. The rapid variations are the
consequence of rotation of the CubeSat around its axis aligned on Earth’s
magnetic field, here set at one degree per second.
Figure 4.11 shows associated ESATAN results. It is obvious that they
are quite different from those exposed in the preliminary analysis. The
bar plot on the right shows the maximal temperature reached at the main
nodes during the orbit. They are all comprise between 30 and 35

C. But,
considering the maximal allowable temperature of the batteries (45

C),
this simulation confirms the result obtained in the preliminary analysis :
the safety margin is narrow and other solutions may be foreseen. A possi-
ble solution could be imagined by analyzing the graph in the bottom left
corner. It shows that the EPS2 PCB is quite cooler than the EPS PCB since
no power is dissipated on EPS2. Actually, according to [33] and [55], the
dissipated energy on EPS2 will be smaller in any case. A possibility could
then to attach the batteries’ PCB to EPS2 PCB instead of EPS PCB. This
solution will be studied in details in the low level model.
The other reasons for which EPS2 is cooler is deduced from the same
graph. It shows that even if half of the collected power is dissipated on the
COM PCB, it remains cooler than the others mainly because it is linked
with two opposite faces (2 and 5) and when the first one is in sunlight,
the other is consequently turned to deep space. The other fact involving
cooler temperatures of COM and EPS2 PCBs is that the conductive links
with the structure is greater than the one linking the OBC to the structure
(the midplane standoff are made of aluminum, nearly ten times more con-
Lionel Jacques
2
nd
Master in Aerospace Engineering
50 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
0 10 20 30 40 50 60 70 80 90 100
0
5
10
15
20
A
b
s
o
r
b
e
d

p
o
w
e
r

[
w
]
Time [min]


Solar
Albedo
Earth IR
Total
Total − electrical
Electrical
Figure 4.10 – Hot case absorbed power
0 20 40 60 80 100
0
10
20
30
40
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


Face 1
Face 2
Face 3
Face 4
Face 5
Face 6
0 20 40 60 80 100
20
25
30
35
40
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


OBC
OBC2
EPS
EPS2
COM
Batteries
0 10 20 30 40 50
Solar Cells 1
Solar Cells 2
Solar Cells 3
Solar Cells 5
Solar Cells 6
Face 1
Face 2
Face 3
Face 4
Face 5
Face 6
OBC PCB
OBC2 PCB
EPS PCB
EPS2 PCB
COM PCB
Batteries
30.12°C
36.52°C
33.74°C
36.55°C
34.93°C
30.17°C
36.18°C
33.64°C
29.98°C
36.2°C
34.82°C
34.78°C
35.59°C
36.63°C
34.76°C
34.15°C
36.32°C
Maximal Temperature [°C]
Figure 4.11 – Hot case temperatures
Lionel Jacques
2
nd
Master in Aerospace Engineering
51 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
ductive that stainless steel used for bottom spacers, as it was indicated in
the table 4.7).
Another intuitive consequence of the PCB configuration is that central
PCBs (EPS, EPS2) are less sensitive to external temperature variations than
OBC and COM PCB. This is intuitive since central PCBs are more insulated
from the external structure.
One must keep in mind that those temperatures are averaged tempera-
tures which are then not able to represent of local hot spot that may occur,
especially on the PCBs.
The graph in the top right corner illustrates the temperature evolution
of the faces. Face 1 and 4 have a different comportment than the others :
this is a consequence of the passive alignment of one axis of the CubeSat
on Earth’s magnetic field. The rotation about this axis is clearly visible
through the fast variations of faces 2,3,5 and 6 while faces 1 and 4 spends
more time in sunlight or shadow.
Globally, one can see that internal components undergo relatively
lower temperature variations than the external structure. Moreover, the
rotation rate considered here is 1 deg/s. ADCS simulations [25] shows
that the CubeSat will certainly undergo higher rotation rates, depending
on the initial condition after being released from the P-POD and on the
moment generated by the deployment of the antennas. Figure 4.12 illus-
trates the consequence of a rotation rate of 5 deg/s. The variations are then
too fast compared to the thermal inertia of the faces and PCBs. These ones
do not undergo anymore fast variations and are almost constant. A steady
state analysis considering the mean incident fluxes would then be suitable
and more computation-time effective, especially for the next model, much
more complex.
0 10 20 30 40 50 60 70 80 90 100
0
5
10
15
20
25
30
35
40
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


Face 1
Face 2
Face 3
Face 4
Face 5
Face 6
0 10 20 30 40 50 60 70 80 90 100
20
25
30
35
40
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


OBC
OBC2
EPS
EPS2
COM
Batteries
Figure 4.12 – Evolution of the temperatures during one orbit for a rotation rate of 5
deg/s
Lionel Jacques
2
nd
Master in Aerospace Engineering
52 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
4.5.2 Cold case
Now that the hot case has been analyzed, it is time to examine the
cold one. Once again, the evolution of absorbed powers along the orbit is
represented at figure 4.13. A first difference with the hot case, excluding
the evident eclipse time, is a higher albedo since the CubeSat get over the
subsolar point
1
. The second feature visible is the presence of two off-peaks
around 10 and 90 minutes. These off-peaks occurs when the face 1 (first
one) or 4 (second one) is perpendicular to sun rays, the projected area
being much lower. The second off-peak is lower than the first one because
of the lower absorptivity of face 4, not covered with solar cells.
0 10 20 30 40 50 60 70 80 90 100
0
5
10
15
20
25
A
b
s
o
r
b
e
d

p
o
w
e
r

[
w
]
Time [min]


Solar
Albedo
Earth IR
Total
Total − electrical
Electrical
Figure 4.13 – Cold case absorbed power
Concerning the temperatures, results are illustrated by the figure 4.14.
The off-peaks visible in the absorbed power are turned here into peaks:
shortly after 10 minutes, the upper left graph shows that face 1 undergoes
a rise of its temperature.
But the important fact is not there: the temperature of the batteries
drops below -20

C while its allowable minimal temperature in charge is
0

C! Obviously, a heater will be needed to raise the temperature of the bat-
teries. But one must pay attention to the limits of the model : as each PCB
is modelled with one node, no transverse conduction is taken into account
in the link between the EPS PCB and the PCB sustaining the batteries.
Furthermore, the assumed material of the spacers between the EPS and
1
As it has been studied in the preliminary analysis, the higher albedo present in the
cold case and not in the hot case does not affect significantly our worst case definition
Lionel Jacques
2
nd
Master in Aerospace Engineering
53 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
0 20 40 60 80 100
−40
−30
−20
−10
0
10
20
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


Face 1
Face 2
Face 3
Face 4
Face 5
Face 6
0 20 40 60 80 100
−35
−30
−25
−20
−15
−10
−5
0
5
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


OBC
OBC2
EPS
EPS2
COM
Batteries
−50 −40 −30 −20 −10 0
Solar Cells 1
Solar Cells 2
Solar Cells 3
Solar Cells 5
Solar Cells 6
Face 1
Face 2
Face 3
Face 4
Face 5
Face 6
OBC PCB
OBC2 PCB
EPS PCB
EPS2 PCB
COM PCB
Batteries
−30.12°C
−28.52°C
−32.14°C
−28.7°C
−28.62°C
−30.09°C
−28.44°C
−32.05°C
−30.15°C
−28.57°C
−28.5°C
−22.03°C
−21.8°C
−21.88°C
−24.47°C
−27.62°C
−21.4°C
Minimal Temperature [°C]
Figure 4.14 – Cold case temperatures
BAT PCB is aluminum which is a good conductor. These two assumptions
are not favorable to the batteries. The influence of the EPS-BAT spacers
conductivity
Again, the COM PCB is cooler than the others for the same reasons
that those previously presented.
4.5.3 Sensitivity analysis
Cold case has shown that the batteries had approximatively the same
temperature as the EPS PCB one. But this was for EPS-BAT spacers in
aluminum. The scope of this sensitivity analysis is then to evaluate the
effect of reducing the thermal conductivity of these spacers.
Figure 4.15 shows that insulating the batteries has a positive effect:
reducing the conductivity from 150 to 5 W/mK increase the minimal tem-
perature reached by the batteries from -21 to -17

C. The graphic on the
right shows that the efficiency of this effect is maximum between 1 and
100 W/mK. Considering the definition of a conductive link GL =
kS
L
,
increasing the length or reducing the area would have the same effect
but structural constraints must be taken into account (resistance, rigidity,
density,. . . ). Again, materials having very low thermal conductivity are
thermoplastics or thermosets but not compatible with structural aspects.
Using CES EduPack, titanium and titanium alloys appeared to be a good
choice and having a thermal conductivity about 10 W/mK with a high
Young’s modulus and high tensile strength.
Lionel Jacques
2
nd
Master in Aerospace Engineering
54 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 4. Simplified Thermal Model
0 20 40 60 80 100
−25
−20
−15
−10
−5
0
Time [min]
B
a
t
t
e
r
i
e
s

t
e
m
p
e
r
a
t
u
r
e

[
°
C
]


k = 0.1 [W/mK]
k = 1 [W/mK]
k = 10 [W/mK]
k = 50 [W/mK]
k = 100 [W/mK]
k = 400 [W/mK]
10
0
10
2
−22
−21
−20
−19
−18
−17
−16
EPS−BAT spacers’ conductivity [W/mK], log scale
B
a
t
t
e
r
i
e
s

m
i
n
i
m
a
l

t
e
m
p
e
r
a
t
u
r
e

[
°
C
]
Figure 4.15 – In uence of the EPS-BAT spacers conductivity on batteries’ temperature
Once again, further analysis will be developed with the low level
model because some important effects neglected here such as transverse
conduction through PCBs will be taken into account.
Nevertheless, a combined solution, both passive, through batteries in-
sulation, and active with small heaters, will certainly be required to main-
tain the batteries in their safe range of temperatures.
Summary
This Simplified Thermal Model has allowed to evaluate the average
temperatures of each PCB and confirmed the potential problems consid-
ering the batteries for both hot and cold cases. However, even if it allowed
to foresee some possible solutions to these problems, several assumptions
has to be removed in order to properly design suitable solutions. This is
the aim of the Detailed Thermal Model. However, before presenting the
Detailed Thermal Model, some measurements are required and discussed
in the next chapter.
Lionel Jacques
2
nd
Master in Aerospace Engineering
55 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
5
Measurements
This chapter describes several measurements performed at the Cen-
tre Spatial de Liège, notably about the batteries. The thermal design of a
satellite involves a good knowledge of thermo-optical and materials prop-
erties, as it has been shown previously. But in a student project, only few
measurements can be performed. In this chapter, several experiments will
be done to determine different properties of the aluminum frame and the
batteries.
5.1 Why and which measurements ?
Li-Po batteries thermal properties are not well known and difficult
to find. Yet, as already spotted, they are the more critical unit seeing
their narrow range of temperatures and measurements seemed necessary.
Moreover, the data used in the previous models were based on Swiss-
Cube model but their batteries are packed in small aluminum boxes. That
is the reason why the first experiment concerns the determination of the
thermal properties (conductivity and the specific heat) of a typical LiPo
battery such as the one probably used in our CubeSat.
The second measurement focuses on the thermal emissivity of the alu-
minum frame’s two coatings : alodine and hard anodizing. This test is
based on a thermographic measurement. This is also the case for the third
one about the thermal contact between the top and bottom faces with the
lateral frame, around the screws.
5.2 The Battery
The goal of this test is thus to evaluate the thermal capacitance and
transverse conductivity. These two properties are crucial when designing
heaters : how much power will be required to maintain the two batteries
above a given threshold and where to put the control sensor. Indeed, the
combination of thermal capacitance and conductivity gives us a delay be-
tween the time the heater is turned on and the sensor response. It acts like
an hysteresis and prevent the heaters from too fast pulsing.
5.2.1 Experimental setup
This test will be performed on a battery Kokam SLPB-554374-H which
is representative since the foreseen models (Varta, Kokam, Panasonic) are
all of approximately the same size and made upon the same technology.
56
Chapter 5. Measurements
The principle of the test is to heat one side of the battery while impos-
ing the temperature of the opposite side. By this way, the temperature of
the heated side will rise up to its constant steady state value. To impose
the temperature of the "rear side", the battery is fixed on an interface plate
which is itself strongly fixed with bolts on a massive steel bench. Seeing
its large capacitance, the temperature of the bench is assumed to remain
constant and equal to the ambient temperature. Thermal compound
1
is
used between the interface plate and the battery to ensure a good contact.
As shown on the figure 5.1, the battery is then strongly pressed on the in-
terface plate with two pieces of CFRP. CFRP has been chosen for its high
rigidity to thermal conductivity ratio. As all the test is performed under
ambient conditions, heaters are covered with aluminum tape to prevent
heat flux from escaping radiatively and all the setup is covered with MLI
2
again to diminish radiative losses and reduced convective exchanges.
Figure 5.1 – Battery test setup
Concerning the heater used is a Minco HK5164R39.2. Unfortunately,
the heater does not cover the entire area of the battery as shown on
the right picture of the figure 5.1 but this will be taken into account in
the model. To reduce errors, multiple consecutive heating cycles are per-
formed and the model will be adjusted for all the cycles. Acquisition is
made through thermocouples and PT100 sensors linked to a PC equipped
with a Keithley 2700 datalogger.
5.2.2 The model
In order to determine the two desired properties, a Simulink
TM
model
fitting at best the setup has been created. Simscape
TM
toolbox available
in Matlab
TM
Simulink
TM
R2008 is used, allowing to create transient ther-
mal model with simple constitutive elements: thermal masses, conduc-
tive/radiative/convective links, heat and temperature sources and sen-
sors etc. . . Our model is shown on the figure 5.2. It contains four thermal
masses : two for the top and bottom side of the battery, one for the inter-
face plate and one for the bench. Indeed, seeing the results, the tempera-
ture of the battery bottom side is no more imposed to model the observed
local heating of the interface plate and bench. This model takes also into
1
Wakefield Engineering Thermal Compound n

126-4
2
Multi Layer Insulation with Dacron netting or "bridal veil" between the layers to reduce
contact
Lionel Jacques
2
nd
Master in Aerospace Engineering
57 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 5. Measurements
account the conduction through the thermal compound between the bot-
tom side and the interface plate.
The same consecutive powers really injected in the heater is set as input
and the three measured temperatures are set in output: battery both sides
and the interface plate.
Out 2
2
Out 1
1
f(x)=0
PS S
Power
t P fcn
PS S
PS S
PS S
Interface Plate
A
B
T
A
B
T
A
B
T
S
A
B
A
B
A
B
A
B
Clock
Bench
Battery Top
Battery Bottom
Figure 5.2 – Battery test Simscape
TM
model
5.2.3 Model adjustment & results
The SLPB-554374 real dimensions differed somehow from the one
mentioned in the datasheet : the real ones were 65 40 5 [mm] (in-
stead of 70 42.5 5.6 [mm]) which leads to a volume of 13000 [mm
3
] for
a weight of 34g.
As the heater does not covered all the area and thanks to the different
sensors
1
used, it has been observed that the temperature of the top side
was of course not uniform. Intuitively, one decided to consider that the
heat flux propagates through a reduced area equal to mean area between
the heater and the bottom side (
65+50
2

40+25
2
[mm]). It is such as the heat
flux propagates according to a truncated pyramid shape whose bottom
face would be the bottom side and top face the heater.
The measured temperatures presented at the figure 5.3 results from an
averaging between the temperatures measured by the different sensors.
The thermal inertia of the interface plate has its importance: it acquires
and releases energy and contributes to the damping of the system. At
the same time, if the bench had not been modelled and replaced by a
fixed temperature, the global heating of the system could not have been
represented (the temperature ad the end of the test is greater than the
initial one because the bench absorbed some energy and did not have
the time to release it seeing its large thermal inertia). By adjusting the
different parameters of the model, the measured temperatures are finally
1
We had 4 sensors on the top side, 2 on the bottom side and 2 on the interface and
bench
Lionel Jacques
2
nd
Master in Aerospace Engineering
58 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 5. Measurements
fitted and the desired properties are determined: the specific heat is about
1350 [J/kgK] and the conductivity 1.55 [W/mK] (through the reduced area
which is equivalent to 1.11 [W/mK] through the entire area).
0 10 20 30 40 50 60 70
20
25
30
35
40
45
50
55
60
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


Top face, measured
Bottom face, measured
Interface plate, measured
The same, simulated
Figure 5.3 – Battery test comparison between measurements and adjusted model
This involves a heat capacity of 45.9 [J/K]. Although the internal com-
position of the battery is difficult to known, it appears that the electrolyte
of Li-Po batteries is held in polymer composites such as polyethyleneoxide
(PEO) or polyacrylonitrile. Such materials have high specific heat about
1800 [J/kgK]. On the other hand, the cathode, anode and packaging shell
have lower specific heat. This leads us to believe that the determined value
is not too far from reality. The results were also compared with data from
specific measurements on other Li-Po batteries [60].
One must keep in mind that the employed method relies on some
assumptions : radiation and convection losses were neglected even though
limiting their effects. The representativeness of the test compared with the
real configuration is also limited since larger heaters will probably be used
to cover all the available area and batteries will not be fixed on aluminum
plate but on CFRP (PCB).
5.3 aluminum frame emissivity
5.3.1 Anodizing vs Alodine
The second test mentioned concerned the measurement of the emis-
sivity of the Pumpkin structure. The frame has two different surface treat-
ments: the rail are hard anodized and the remaining surface is alodyned
and here are a few words about these two processes. The purpose of both
coatings is to improve corrosion resistance but while Alodine maintain
electrical conductivity, anodized surfaces have a enhanced protective ox-
Lionel Jacques
2
nd
Master in Aerospace Engineering
59 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 5. Measurements
ide layer which is also a poor electrical conductor. Anodizing is an elec-
trolytic passivation process increasing the oxide layer thickness. Alodine is
a chemical conversion coating (also known as Iridite or Chromate Conver-
sion). Both treatments improve adhesion for paint primers and glues. In
our case, the rail were anodized to prevent galling (as a reminder, the rails
are the only part of the CubeSat in contact with he P-POD). According to
Adam Reif from Pumpkin, here are the military specifications (MIL-spec)
reference of these two processes: MIL-A-8625F, TYPE III, CLASS 1 [38]
for the anodized process and MIL-DTL-5541F Type II, CLASS 1A for Alo-
dine [39]. Type III CLASS 1 means non-dyed hard anodizing and Type II,
CLASS 1A for alodine means that no hexavalent chromium (environment
friendly) is used in the process (Type II) and that maximum protection
against corrosion is provided (class 1A).
Nevertheless, precise data about alodine and hard anodized emissiv-
ity were quite difficult to find and optical properties are highly surface
treatment dependent. For example, figure 5.4 excerpt from a NASA report
about thermal control coatings, shows how the chemical reaction time of
a chemical conversion coating on aluminum affects its optical properties.
Figure 5.4 – Evolution of the emissivity and absorptivity of aluminum sample subjected
to a conversion coating surface treatment in function of the reaction time [61]
So, emissivity measurements were planned. Many thermal emissivity
measurements method already exist but they often required specific se-
tups and expensive apparatus. A simple and relatively convenient way of
doing this is to use a thermographic camera.
5.3.2 Measuring emissivity with a thermographic camera
Here is a brief theoretical recall about thermographic measurements.
As depicted in Figure 5.5, one must be very cautious on how to interpret
the measured data. Indeed, the total input power collected by the camera,
W
tot
, has different contributions:
1. emission from the object we are looking at;
Lionel Jacques
2
nd
Master in Aerospace Engineering
60 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 5. Measurements
2. reflected emission from ambient sources;
3. atmosphere emission.
15 The measurement formula
As already mentioned, when viewing an object, the camera receives radiation not
only from the object itself. It also collects radiation from the surroundings reflected
via the object surface. Both these radiation contributions become attenuated to some
extent by the atmosphere in the measurement path. To this comes a third radiation
contribution from the atmosphere itself.
This description of the measurement situation, as illustrated in the figure below, is so
far a fairly true description of the real conditions. What has been neglected could for
instance be sun light scattering in the atmosphere or stray radiation from intense ra-
diation sources outside the field of view. Such disturbances are difficult to quantify,
however, in most cases they are fortunately small enough to be neglected. In case
they are not negligible, the measurement configuration is likely to be such that the
risk for disturbance is obvious, at least to a trained operator. It is then his responsibil-
ity to modify the measurement situation to avoid the disturbance e.g. by changing
the viewing direction, shielding off intense radiation sources etc.
Accepting the description above, we can use the figure below to derive a formula for
the calculation of the object temperature from the calibrated camera output.
10400503;a1
Figure 15.1 A schematic representation of the general thermographic measurement situation.1: Surround-
ings; 2: Object; 3: Atmosphere; 4: Camera
Assume that the received radiation power Wfroma blackbody source of temperature
T
source
on short distance generates a camera output signal U
source
that is proportional
to the power input (power linear camera). We can then write (Equation 1):
15
Publ. No. 1 558 071 Rev. a196 – ENGLISH (EN) – December 21, 2006 125
Figure 5.5 – A schematic representation of the general thermographic measurement
situation from the ThermaCAM User Manual [21]. 1: Surroundings; 2: Object; 3:
Atmosphere; 4: Camera
These contributions involve four main parameters the user has to sup-
ply:
the object emissivity
the atmosphere transmittance τ depending on the distance between
the object and the camera, the atmosphere relative humidity and its
temperature
surroundings average reflected temperature
surroundings effective emissivity
Concerning, the surrounding effective emissivity, all radiation emitted
from the object is assumed to never come back so that the environment
can be considered as a blackbody and its effective emissivity is one.
The total collected radiation power seen by the camera is then :
W
tot
= τW
obj
+ (1 )τW
re f l
+ (1 τ)W
atm
and the radiation power effectively emitted by the object is given by:
W
obj
=
1
τ
W
tot

1

W
re f l

1 τ
τ
W
atm
These equations show that to determine the effective object emitted
power, the camera has to subtract the two unwanted contributions. But as
the emissivity decrease, the contribution coming from the reflected tem-
perature of the surroundings increases according to
1

and so does the
sensitivity to reflected environment parameters. A way to reduce this sen-
sitivity is to increase the ratio between object and surroundings emitted
radiation. The emitted radiation is proportional to the temperature at the
fourth power and the ratio is thus equal to
1

T
4
re f l
T
4
obj
. The influence of these
Lionel Jacques
2
nd
Master in Aerospace Engineering
61 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 5. Measurements
two parameters are represented in Figure 5.6. The lower the emissivity,
the more important is the influence of temperature. For example, with an
emissivity of 0.1, heating the object up to 80

C reduces the ratio from 9 to
4.
20 40 60 80 100 120 140 160 180 200
0
1
2
3
4
5
6
7
8
9
Object temperature T
obj
[°C]
1

T
4
r
e
f
l
T
4
o
b
j


ε=0.1
ε=0.2
ε=0.3
ε=0.4
ε=0.5
ε=0.6
ε=0.7
ε=0.8
ε=0.9
Figure 5.6 – In uence of the object temperature and emissivity for a thermographic
measurement
For this reason, several measurements shall be performed at different
temperatures. In practice, for each temperature, the object emissivity pa-
rameter in the software shall be varied, noticing the resulting temperature
given. If the temperature calculated by the software is equal to the one
measured by the sensor then the corresponding emissivity must be close
to the real one.
5.3.3 Experimental setup
The camera used was a ThermaCAM
R
S40 from FLIR Systems
equipped with a 320 240 microbolometer array.
In order to have a uniform surrounding environment, we enclose the
structure with a high emissivity box of known temperature. We placed the
camera close to the structure to limit atmosphere effects (Relative humid-
ity was 47% and distance 30cm) and tilted the frame to avoid camera’s
own emitted radiations reflection like looking in a mirror (figure 5.7, left).
The structure was heated through a plate on which we fixed heaters, as
shown the the right picture of the figure 5.7.
Figure 5.8 shows the infrared picture of the structure heated up to
90

C. The two considered areas for the emissivity calculation are the one
enclosed by green rectangles, close to which sensors are set as shown in
figures 5.7 and 5.8.
Lionel Jacques
2
nd
Master in Aerospace Engineering
62 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 5. Measurements
Figure 5.7 – aluminum frame emissivity test setup
Figure 5.8 – Infrared capture of the structure heated up to 90

C
5.3.4 Results
First, the method has been validated by measuring samples whose
emissivity were well known : 3 mil kapton foil on aluminum backing
( = 0.81, [52]), common aluminum plate (

= 0.1, [21]) and white pa-
per sheet (

= 0.8, [21]).
Then, as previously said, measurements were performed at different
temperatures and for each of them, the emissivity is varied and a rela-
tive error can be deduced between the measured value and the infrared
calculated one: T
IR
T
mes
T
mes
The evolution of this error in function of the measured temperature
and emissivity is represented in Figure 5.9. The left plot represents the
hard anodized results while the right plot displays the Alodine ones.
The Alodine curves are more emissivity sensitive and that confirms
the theoretical curves of Figure 5.6 showing that when the emissivity was
low, two consecutive curves were well spaced. Figure 5.9 also confirm that
as the measurement temperature increase, the emissivity leading to a zero
error seems well to converge.
Seeing these graphs, we can finally conclude that the emissivity of the
Alodine and hard anodized coatings are respectively about 0.14 and 0.69.
Lionel Jacques
2
nd
Master in Aerospace Engineering
63 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 5. Measurements
0.65 0.7 0.75
−5
0
5
Hard Anodized rail emissivity
E
r
r
o
r

b
e
t
w
e
e
n

s
e
n
s
o
r

T
°

a
n
d

I
R

m
e
a
s
u
r
e
d

T
°

[
%
]


T = 42.5°C
T = 52.3°C
T = 58.7°C
T = 88.1°C
T = 99°C
0.1 0.12 0.14 0.16 0.18 0.2
−5
0
5
Alodine emissivity
E
r
r
o
r

b
e
t
w
e
e
n

s
e
n
s
o
r

T
°

a
n
d

I
R

m
e
a
s
u
r
e
d

T
°

[
%
]


T = 46.4°C
T = 58.3°C
T = 66.2°C
T = 102.8°C
T = 115.9°C
40 60 80 100
0.64
0.65
0.66
0.67
0.68
0.69
0.7
H
a
r
d

A
n
o
d
i
z
e
d

r
a
i
l

e
m
i
s
s
i
v
i
t
y
Temperature [°C]
40 60 80 100 120
0.11
0.12
0.13
0.14
0.15
A
l
o
d
i
n
e

e
m
i
s
s
i
v
i
t
y
Temperature [°C]
Figure 5.9 – aluminum frame emissivity test results
5.4 aluminum frame contact resistance
The third experiment concerns contact between top and bottom face
screwed on the lateral frame. The contact phenomenon between frame el-
ements is clearly highlighted in Figure 5.10 which presents some screwed-
joints in front of a bright background: only a fraction of the surface area
contributes to heat flow path between the faces.
Figure 5.10 – Aluminum frame contacts in front of bright background. On the left, face
6 and 4, fastened with 3 screws and on the right, face 6 and 1, fastened with one screw
The experiment is done under ambient conditions so that convective
losses are absolutely not negligible and difficult to take into account. Then,
the aim of this experiment is not to compute accurate thermal conduc-
tances values but rather to evaluate the importance of the phenomenon.
For this purpose, infrared imaging is again used.
Lionel Jacques
2
nd
Master in Aerospace Engineering
64 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 5. Measurements
The method consists in imposing a heat flux between faces 3 and 6
through the contact. To do this, face 6 is heated up while face 3 is strongly
fastened to a large bench (high thermal inertia) to prevent its tempera-
ture from rising. By this way, all the heat (except convective and radiative
losses) must flows through the contact resistance. IR imaging can then
show if the contact area is strongly localized or well spread across all the
area of the bracket. Figure 5.11 shows the IR pictures of the contact be-
tween faces 6 and 1 (left) and between faces 6 and 2(right). The first one
confirms that the contact is well localized: the area below the screw is
hotter meaning that only this part of the surface contributes to the con-
tact resistance and that constriction phenomenon occurs (shrinkage and
concentration of the heat flow lines around the contact area). On the con-
trary, the right picture is less convincing: heat flows seems to be more
distributed on all the surface.
Figure 5.11 – Infrared images Battery test setup
In the same way, the other contact areas were also studied leading to
the same conclusions: contact is present but difficult to estimate and varies
from one screwed-joint to another.
Lionel Jacques
2
nd
Master in Aerospace Engineering
65 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 5. Measurements
Summary
The preliminary analysis and simplified thermal model showed that
thermo-optical properties were important and that the batteries thermal
inertia contributes largely to the global thermal of the CubeSat. Measure-
ments allowed us to determine relatively accurate values for the emissiv-
ity of the aluminum frame and for the thermal properties of the battery.
For instance, the emissivity of the frame rails is in reality lower than the
assumed values : 0.7 instead of 0.8 and the heat capacity of the battery
is quite greater: 1350 [J/kgK] instead of 960 [J/kgK]. These new values
will be used in the newt chapter describing the detailed thermal model
of OUFTI-1. Even if the third experiment did not lead to concrete values,
it demonstrated the presence of contact phenomenon between the lateral
frame and bottom/top faces. This phenomenon will then be taken into
account through data found in the literature and finite elements analysis.
Lionel Jacques
2
nd
Master in Aerospace Engineering
66 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
6
Detailed Thermal Model
Introduction
Up to this point, two models have been developed. Both revealed the
need of an active thermal control to keep the temperature of the batteries
above its lower limit. But the Simplified Thermal Model (STM) was too
limited to allow a convenient and realistic design of a heater.
As already pointed, the STM was also unable to represent the potential
hot spots due to low efficiency electronic components such as the ampli-
fier. Moreover, as the model detail evolves along with the design maturity
of the CubeSat, especially electronic circuits, the dissipation system intro-
duced in the previous chapter also evolved. The new design is composed
of a dissipation transistor in series with a resistance instead of resistances
alone. This means that the heat load will probably be strongly localized.
All these issues suggest to develop a more precise model allowing
concrete decisions and solutions. This is the aim of the present chapter
presenting the Detailed Thermal Model (DTM). It is divided into four
main parts : developing, checking, analyzing and optimizing the model.
6.1 Geometric Mathematical Model
Like the Simplified Thermal model and any other thermal model, the
DTM consists of a GMM, used for the computation of the radiative ex-
change factors and environmental absorbed powers, and a TMM, mod-
elling the conductive network between the nodes, their thermal inertia
and dealing with the internal dissipated power distribution.
6.1.1 Nodal breakdown
aluminum Frame
In the Simplified Thermal model, each face of the aluminum frame was
considered isothermal. But as represented in Figure 6.1, the midplane
standoffs, linking the aluminum frame with the COM PCB, are connected
at the top of faces 2 and 5, near the face 3. This was thus not taken into
account in the STM. For this reason, each face of the aluminum frame is
now into several elements.
Actually, to model at best the point where the midplane standoff is
fastened, the face is divided into nine nodes, unequally distributed : the
67
Chapter 6. Detailed Thermal Model
Figure 6.1 – Midplane standoffs fastening to the aluminum frame
Figure 6.2 – Detailed Thermal Model nodal breakdown of the aluminum frame
Lionel Jacques
2
nd
Master in Aerospace Engineering
68 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 6. Detailed Thermal Model
nodes at the corners are smaller. The boundary between the corners and
adjacent nodes is defined by the notch at the extremities of the folds of the
faces 3 and 6, where it is narrower.
Figure 6.2 shows the nodal breakdown of the frame. The complete
numbering convention is available in appendix ??.
CubeSat’s feets, visible in figure 6.1, are also modelled and included in
nodes 3301, 3303, 3305 and 3307 for the upper face and 3601, 3603, 3605
and 3607 for the bottom face.
aluminum panels
To be consistent with the division of the frame, the aluminum panels are
as well divided but in nine equal elements. Equal elements imply easier
computation of the conductive links within the panel but more difficulties
for the transverse links with the frame as this will be exposed later.
Solar cells
In the same way, each solar cell has now its own node. This model takes
into account the cell efficiency variation but this will be treated in the
TMM.
PCBs
Concerning the PCBs, their level of discretization depends on their com-
ponents: high localized dissipative components requires a finer discretiza-
tion. Another criterion was already spotted when considering the link
with the secondary PCB sustaining the batteries : the transverse conduc-
tion within the PCB from main spacers at the corners has to be modeled.
And here are the level of discretization of each PCB, based on these
two criteria:
- OBC & OBC2 : few nodes required because of low power, low dissi-
pation ICs (integrated circuits) 2 2 elements.
- EPS : more nodes required because of transverse conduction model-
ing and high power, high dissipation ICs 10 10 elements such
as the elements are about 1cm wide, approximatively the size of the
dissipation transistor[51].
- EPS2 : because the case of hanging the battery PCB on EPS2 will
be studied, more nodes are also required for transverse conduction
modeling, even if it contains lower power ICs 10 10.
- COM : the only critical and high dissipation IC known at the moment
is the amplifier. Nevertheless, the dissipated power level is such that
a finer discretization is also required 10 10.
In this model, the connectors are modeled and added to the corre-
sponding node as well as the spacers.
In practice, Matlab routines have been developed to make the dis-
cretization of each PCB entirely automatic and autonomous (both for the
GMM and TMM) with only one input: the number of elements desired.
Lionel Jacques
2
nd
Master in Aerospace Engineering
69 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
Antennas
This model includes the two antennas: the 17 cm and the 50 cm one. Once
again, they are divided into several elements. Both antennas are 0.25 mm
thick and 3.5 mm wide and are made of copper [63]. Since the antenna
deployment mechanism panel geometry was not yet clearly defined when
the model was created, it is only modelled with two nodes: one for the
exterior and one for the internal side.
6.1.2 Thermo-optical properties
Thermo-optical properties are mainly the same as the ones used for the
Simplified Thermal Model. The differences comes from the new elements
taken into account :
- the emissivity of the batteries, set as parameter : the effect of covering
the batteries with a low emissive tape will be investigated;
- the emissivity of the rails : the measured value of 0.7 (see previous
chapter) is now used.
- antennas (copper) thermo-optical properties are: = 0.03 and α =
0.3 [24],[50]. Seeing their high α/ ratio, antennas will probably be
relatively hot.
- spacers (made of 6061 aluminum, machine finish) thermo-optical
properties are set to = 0.05 ρ
d,IR
= 0.15 ρ
s,IR
= 0.8.
6.1.3 ESARAD considerations
The nodal breakdown defined above suggests us to split up the
ESARAD geometry in 3 main parts :
- an external skin comprising the solar cells, the aluminum panels,
the antennas and the remaining visible aluminum frame (mainly the
rails and the feet);
- an internal skin representing the environment seen by the PCBs : the
aluminum panels and frame;
- the PCBs stack.
Figure 6.3 illustrates the GMM in ESARAD. As shown by the lower left
picture, the particular geometry of the frame has been modelled. In this
way, the aluminum panels and frame radiative coupling are consistent
with reality.
6.2 Thermal Mathematical Model
6.2.1 Conductive network
aluminum frame
All the conductances of the frame have been determined through finite
element analysis. The detailed procedure is available in appendix B.
Lionel Jacques
2
nd
Master in Aerospace Engineering
70 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
Figure 6.3 – Geometric Mathematical Model in ESARAD
Measurements confirmed the presence of contact phenomenon be-
tween the lateral frame and faces 3 and 6. Data about screwed-joints are
difficult to find because they are often developed within companies and
seldom published. The table of figure 6.4, coming from Spacecraft Thermal
Control Handbook by D. Gilmore, shows contact resistance data
1
for various
bolt size and plate thickness.
Figure 6.4 – TRW and Lockheed Martin Bolted-Joint Resistance Data [24]
The fastening screws used are NC 4-40 (corresponding to M3 metric
standard) and give, for thin plates (1.57 mm), a 12.6 [K/W] contact resis-
tance.
1
from TRW and Lockheed Martin
Lionel Jacques
2
nd
Master in Aerospace Engineering
71 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
Aluminum panels
There are nine nodes per panel, except for the antenna deployment mecha-
nism panel, equally distributed in both directions. The three lateral panels
(faces 1, 2 & 5) are identical, as the top and bottom panels (faces 3 & 6).
Horizontal/vertical conductances (respectively R
H
and R
V
) determination
is then straightforward:
R
H
=
w
k (th)
R
V
=
h
k (tw)
where w = W
panels
/3, h = H
panels
/3 and k the thermal conductivity of
the 7075 aluminum panel.
Transverse links
The thermal modelling of transverse links between the frame and the pan-
els and between the panels and the solar cells is relatively complex, mainly
because the nodes are not directly opposed due to their different cross sec-
tions. But as the different layers are thin, one can assume that the flux lines
remains perpendicular to the cross section, neglecting their extension (and
thus the increase of the local cross section) in the larger layer. Therefore,
the surface area involved in the thermal conductance formula (GL =
kS
L
) is
taken as the smallest cross section of the two nodes. For example, consid-
ering the case of the figure 6.5 where the heat flux flows from the upper
plate to the lower one, the surface S would be equal to S
1
and the length
L = (t
1
+ t
2
)/2 (the nodes are set in the middle of each plate thickness).

1

2

1
+
2

Figure 6.5 – Thermal conductance between two thin plates of different section
In practice, as the integration layouts (solar cells/panels and
frame/panels) are the same for all the nodes, the transverse conduc-
tance is calculated according to the figure 6.6:
1
GL
ij
= R
ij
=
1
S
e f f ective
_
t
cell
k
cell
+
t
adhesive
k
adhesive
+
t
Kapton
k
Kapton
+
t
adhesive
k
adhesive
+
t
panel
2k
panel
_
for the solar cells/panels links and
1
GL
ij
= R
ij
=
1
S
e f f ective
_
t
panel
2k
panel
+
t
adhesive
k
adhesive
+
t
f rame
2k
f rame
_
for the frame/panels links where S
e f f ective
is the common area between
nodes i and j, as defined above, and is computed through CATIA V5 CAD
software.
Lionel Jacques
2
nd
Master in Aerospace Engineering
72 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model

150 Solar cell
RTV S691 adhesive
7075 Aluminium
panel
5052 Aluminium
frame
Adhesive ?
RTV S691 adhesive
2mil Kapton
®

80
50
80
1.5
?
1.27
1.52
or
Figure 6.6 – external layout and corresponding thermal conductance
The adhesive thickness between the frame and the panels is not yet
defined and the same values as the other adhesive layers are assumed.
Inner PCBs links
Now, the PCBs are divided into several elements, to take into account
conduction inside the PCBs.
GL
x
=
k
PCB
_
t
PC104
l
y
_
l
x
=
k
PCB
(t
PC104
L
PC104
)
W
PC104
GL
y
=
k
PCB
(t
PC104
l
x
)
l
y
=
k
PCB
(t
PC104
W
PC104
)
L
PC104
where l
x
=
W
PC104
N
and l
y
=
L
PC104
N
with N the number of elements in
which the PCB is divided in both directions.
Inter PCBs links
Like in the STM, both spacers and bus conductances are taken into ac-
count. The drill holes at the corners of the PCBs consist of copper washers
for heat rejection and structural purposes as depicted in figure 6.7.
The contact between the spacers and the washers is only ensured by
the vertical pressure applied at the bottom of each endless screw, at the
midplane standoffs. A contact resistance is thus added between the spacer
and the PCB washer. The copper being a good thermal conductor (k 400
[W/mK]), conduction through the washer is neglected, compared to the
other resistances. Figure 6.7 describes the inter PCBs conductive model.
Contact resistance is also added between the midplane standoff and the
aluminum frame. OBC PCB is supported by steel inserted threaded spac-
ers into which endless screws are fastened.
Conduction through the endless screws is neglected (large clearance
with spacers and washers) but their heat capacity is distributed on the 4
corresponding spacers.
Lionel Jacques
2
nd
Master in Aerospace Engineering
73 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model





Face 6 corner node
Face 2 or 5 corner
node










+1


3 ×


4




1




6

Figure 6.7 – On the left, inserted copper washers in EPS2 engineering model. On the
right, the inter-PCBs conductive model
Concerning the connector, only the pins contributes to the conduction
and perfect contact is assumed between the pins of consecutive connectors.
EPS PCB to battery PCB links
The way the BAT PCB is fastened to the other PCB was considered as
aluminum spacers in the STM, with perfect contact. STM showed that in-
sulating the BAT PBC by using spacers made of materials less conductive
would be worthwhile. On another hand, STM also showed that the pos-
sibility of fastening the BAT PCB on the EPS2 instead of EPS should be
investigated. Therefore, the conductance between the BAT PCP and the
EPS or EPS2 PCB is kept as a parameter. The parameter who vary is the
conductivity of the spacer, its geometry remaining unchanged. The default
value corresponds to aluminum spacers.
Antennas
Antennas being divided into several elements, each node is linked to the
previous and next node with the same conductance. The contact between
both antennas and the panel is assumed to be perfect. The antennas are
also linked to the amplifier through a coaxial wire. The corresponding
conductance consist of a 5 cm long copper wire with a 1 mm diameter for
both antennas.
6.2.2 Internal power
This section is devoted to the description of the internal power dis-
tribution which only concerns the hot case. But before distributing the
Lionel Jacques
2
nd
Master in Aerospace Engineering
74 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
power, it must be collected: no power can be created out of nowhere and
the section begins thus with the description of the solar cells.
Solar cells ef ciency
The way the electrical power is subtracted has already been exposed in the
STM. But as previously introduced, this model takes into account the tem-
perature dependence of the solar cells efficiency along with their degra-
dation over time. The efficiency is assumed to vary linearly with tempera-
ture
1
according to the following law:
η(T) = η
0
+
∂η
∂T
(T 28

C)
where η
0
is the nominal efficiency at 28

C. However, data about the
new solar cells are not yet available and the temperature coefficient is
based on the 28% cells data, summed up in table 6.1 :
∂η
∂T
=

∂T
_
P
max
C
S
A
SC
_
=
1
C
S
A
SC
_
V
max
∂I
max
∂T
+ I
max
∂V
max
∂T
_
where A
SC
is the cell area and C
S
the solar constant
BOL 2.5E14 5E14 1E15
aging factors [-]
Average efficiency η
0
[%] 28 0.95 0.92 0.86
Voltage at P
max
, V
MP
[mV] 2371 0.96 0.94 0.93
Current at P
max
, I
MP
[mA] 487 0.99 0.97 0.94
dV
MP
/dT [mV/

C] -6.1 -6.8 -6.3 -6.4
dI
MP
/dT [mA/

C] 0.28 0.36 0.2 0.29
Table 6.1 – 28% solar cells efficiency parameters [7], for C
S
= 1367 [W/m
2
]
The last three columns represent the effect of aging due to the radi-
ation environment: the cells efficiency deteriorates over time because the
space radiation environment affects the materials used in the cells. For
instance, the coverglass transmittance decrease with the absorbed dose.
The column headers are the fluence of 1MeV electrons received by the
cell expressed in [Electrons/cm
2
]. In those columns are the coefficients by
which the efficiency, V
MP
and I
MP
must be multiplied, representing the
deterioration over time. Concerning dV
MP
/dT and dI
MP
/dT, the values
does not decrease monotonically over time and their values in function of
the fluence are directly available without the use of an aging coefficient.
Spenvis simulations show that, over our mission duration of one year, the
fluence is 4.3E11 while over the predicted lifetime ( 4.8 years), it is 2E12.
However, since our solar cells are new ones with new aging coefficients
still unknown and since they have not yet been tested in real space condi-
tions the worst case value are used for the hot case. Figure 6.8 shows both
effects of temperature and aging on the solar cells efficiency. The effect of
deterioration is mostly visible on the nominal value of the efficiency but
1
Note that the linear temperature dependence must surely be valid only in a given
temperature range.
Lionel Jacques
2
nd
Master in Aerospace Engineering
75 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
the effect of temperature is clear : the cooler the solar cell, the higher its
efficiency.
−60 −40 −20 0 20 40 60 80
0.22
0.24
0.26
0.28
0.3
0.32
0.34
0.36
Temperature [°C]
C
e
ll
e
f
f
ic
ie
n
c
y

η

[
%
]


BOL
EOL 2.5E14
EOL 5E14
EOL 1E15
Figure 6.8 – Effect of temperature and aging on solar cells efficiency
Internal power distribution
Once the electrical power collected, a fraction is dissipated into heat be-
cause the electronic components are not perfect and have their own effi-
ciency. However only the more dissipative and less efficient components
are modelled: the dissipation system and the amplifier. At the time the
STM was being developed, the dissipation system was composed of four
resistances dispatched at the corner of the EPS PCB. Since that time, it has
evolved and consists now in one transistor. The position of both elements
is not defined and so subjected to optimization.
The cold case assumptions are identical to the one of the STM: all
the collected energy is stored under sunlight with no internal dissipation
throughout the orbit.
The hot case is somewhat different: now that a more detailed power
budget is available and that a more accurate distribution can be realized,
two different hot cases are considered. The first one assumes that the
CubeSat is nearly turned off and that, the batteries being fully charged,
all the collected power is dissipated through the transistor of the dissipa-
tion system.
The second one assumes that D-STAR is turned on involving a full uti-
lization of the amplifier. According to the COM subsystem [36] and [28]
and based on the link budget developed by MIAS [9], the power required
on the antennas is about 750 mW. Assuming that the efficiency of the am-
plifier, not yet determined, is at worst 30%, this means that the amplifier
dissipates 1.75 W into heat. Following back the power path, the ampli-
fier input power (0.75/0.3 = 2.5 W) is provided by the converter whose
efficiency is about 80%. Once again, this means that a part of the input
power of the converter is converted into heat (2.5/0.8 2.5 = 625mW).
OBC and OBC2 have relatively low power consumption and only 2 0.05
Lionel Jacques
2
nd
Master in Aerospace Engineering
76 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
W are uniformly distributed on the two PCBs. According to [33], the ino-
vative EPS dissipation is about 0.3 W and the remaining power is finally
dissipated through the dissipation system.
Hot Case 1 Hot Case 2
P
OBC
[W] 0 0.05 uniform
P
OBC2
[W] 0 0.05 uniform
P
EPS
[W] P
tot
diss. tansist. 0.625 +P
rem
uniform
P
EPS2
[W] 0 0.3
P
COM
[W] 0 1.75 amplifier
Table 6.2 – DTM Hot Cases internal dissipation
By default, both amplifier and dissipation transistor are located at the
center of their respective PCB
6.3 Checking the model
This section is devoted to the check of both GMM and TMM.
6.3.1 GMM
As for the STM, the number of rays fired by the MCRT method has
been selected after a convergence analysis for different seeds but is not
represented here. In addition to that, an analytical estimation of the REF
between two PCBs is carried out using Gebhart’s theory and then com-
pared to ESARAD results. Gebhart factors B
ij
take into account multiple
reflections and depend no longer only on the geometry like the view fac-
tors but also on the thermo-optical properties of the different surfaces. REF
are directly obtained from the Gebhart factors B
ij
thanks to the following
relation:
GR
ij
=
i
A
i
B
ij
The analytical model includes two PCBs facing each other and divided
in 10 10 elements. The bus and the spacers are neglected. The radiative
coupling with the aluminum structure is modelled by closing the assembly
with aluminum sides joining PCB perimeter.
Gebhart factors B
ij
are defined by the equation:
B
ij
= F
ij

j
+

k
F
ik
(1
k
)B
kj
or, in matrix form:
B = β + (F β).B β
ij
= F
ij

j
Details about the view factors computation is available in appendix C.
Figure 6.9 compares ESARAD MCRT and analytical results for one
node of EPS2 PCB (16054) viewing COM PCB nodes (EPS2 and COM
have the same number of nodes). The maximum REF is obviously with
the opposite node (GR
16054,18054
). The main differences occur at the border
Lionel Jacques
2
nd
Master in Aerospace Engineering
77 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
because the analytical model does not take into account the connector,
the gap between the PCB and the aluminum structure and the spacers.
The REF with the corners is indeed lower because the node does not see
them directly and the corners have an area reduced by the presence of the
spacer. Apart these local differences, both surfaces are relatively identical
and confirm that the number of rays fired is sufficient and that the REF
computation through MCRT method has converged.
Figure 6.9 – ESARAD radiative coupling check for one PCB node (left: analytical,
right: MCRT)
6.3.2 TMM
Moreover, being much more complex than the STM, checking the con-
ductive network seemed necessary. For this purpose, three tests are carried
out about different parts of the global conductive network: the lateral alu-
minum frame, the PCB inner links and the links between PCBs.
Aluminum frame
In order to check the lateral aluminum frame conduction, the temperature
is prescribed on face 4 while power is injected on the opposite face. The
temperature reached is then compared with a finite element model under
the same conditions. The comparison is presented in figure 6.10
1
. It con-
firms that the way the nodes are distributed and connected is consistent
with the accurate temperature distribution of the FEM.
PCBs
The PCB discretization is critical and therefore have to be tested. Three
checks are indeed carried out to verify the good implementation of the
conductive network. The first one concerns the verification of the links
between the nodes inside each PCB. It is achieved by imposing a gradient
across one diagonal of the PCB by prescribing the temperature of two
opposite corners. By this way, one can verify the shape of the isotherms
normally perpendicular to the heat flow. Figure ?? shows the results of
1
For the sake of clarity, face 2 is undisplayed in ESARAD view.
Lionel Jacques
2
nd
Master in Aerospace Engineering
78 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
Figure 6.10 – check
this test. The shape of the isotherms are well as expected: nearly (because
of the rectangular shape of the PCB) perpendicular to the diagonal of the
PCB.
Figure 6.11 – Checking the conductive links inside PCBs through isotherm shape
The second test concerns the links through the connectors. The auto-
matic implementation is quite difficult because the number of elements
of two consecutive PCBs may be different (for instance, the links between
OBC2-EPS). The way these links are tested is the following one: power is
injected on EPS PCB while both OBC and COM have a prescribed tem-
perature at their extremity. The temperature reached where the power is
injected is then compared to an simple conductive network, depicted in
figure 6.12. The temperature reached at the end of the EPS PCB, T
2
, is
obtained thanks to the following relation, deduced from the resistance
model:
T
2
= T
1
+ Q
_
R
EPS
+
(R
OBC
+ R
s1
+ R
s2
) (R
COM
+ R
s3
+ R
s4
)
R
OBC
+ R
s1
+ R
s2
+ R
COM
+ R
s3
+ R
s4
_
Lionel Jacques
2
nd
Master in Aerospace Engineering
79 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model


4

3

2

1

2

1

1


Figure 6.12 – Check of the conductive links through the bus and equivalent resistance
model
where
R
EPS
= R
COM
=
L
PC104
n
n 1
W
BUS
/2
k
PCB
t
PC104
n = 10
R
OBC
=
L
PC104
n
n 1
k
PCB
t
PC104
n = 2
R
s1
= R
s2
= R
s4
=
15 mm
k
spacer
A
spacer
R
s2
=
25 mm
k
spacer
A
spacer
For T
1
= 0

C and Q = 5W one gets T
2
= 261.2

C while ESATAN
gives 259

C.
The global convergence of the TMM has been analyzed through the
different convergence parameters ()
6.4 Results
Now that the DTM is checked, concrete simulations can be launched
to analyze this initial design.
6.4.1 Absorbed powers
Figures 6.13 and 6.14 show the different contributions to the total
power absorbed by the satellite during one orbit in the cold and hot cases,
respectively. They are nearly identical to those of the STM except that
slightly more power (for instance, 20.2 W instead of 19.8 W in the hot
case) is absorbed because of the antennas and feet contributions. Never-
theless, important differences will be observed in the temperatures thanks
to the better accuracy of this thermal model.
Lionel Jacques
2
nd
Master in Aerospace Engineering
80 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
0 10 20 30 40 50 60 70 80 90 100
0
5
10
15
20
25
30
Time [min]
A
b
s
o
r
b
e
d

p
o
w
e
r

[
W
]


Solar
Albedo
Earth IR
Total
Total − Electrical
Electrical
Figure 6.13 – Cold case absorbed power
0 10 20 30 40 50 60 70 80 90 100
0
5
10
15
20
25
Time [min]
A
b
s
o
r
b
e
d

p
o
w
e
r

[
W
]


Solar
Albedo
Earth IR
Total
Total − Electrical
Electrical
Figure 6.14 – Hot case absorbed power
Lionel Jacques
2
nd
Master in Aerospace Engineering
81 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
6.4.2 Cold Case
Since the STM showed that the most critical components in the cold
case were the batteries, they are the only internal part represented in Fig-
ure ?? showing also the temperature evolution and solar cells. The first key
difference with the STM is clearly visible: if one reminds that the temper-
ature of the batteries dropped to -21

C in the STM, it drops now only to
-15

C. This translates the effects of transverse conduction within the PCBs
(here the EPS) and contact with spacers both increasing the total thermal
resistance between the batteries and the structure.
0 10 20 30 40 50 60 70 80 90 100
−40
−35
−30
−25
−20
−15
−10
−5
0
5
10
Time [min]
T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


Solar Cells Face 1
Solar Cells Face 2
Solar Cells Face 3
Solar Cells Face 5
Solar Cells Face 6
Antenna deployment mechanism panel
Battery
Figure 6.15 – Cold case absorbed power
Yet, this temperature is still too cold and heaters are required. But
before beginning the design of the heater, the two hot cases have also to
be considered to keep a global point of view and allow a global design.
6.4.3 Hot Case 1
The STM showed that a steady-state analysis using the mean absorbed
powers would possibly be equivalent and much more time effective than a
transient computation. Indeed, as expected viewing the similarities in the
absorbed powers of STM and DTM, a comparison with transient analysis
confirmed that the steady-state one was equivalent and that the fast vari-
ations of the absorbed powers does not influence the mean results. From
now, steady-state analysis are then performed for the hot case instead of
transient, requiring much more cpu time.
Lionel Jacques
2
nd
Master in Aerospace Engineering
82 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
As described in the table 6.2, the first hot case assumes that all the
collected power is dissipated through the dedicated transistor.
While the heat flow path are not critical in the cold case because there
are no internal dissipation and because the structure does not undergo
large thermal gradients, a heat flow map is more useful in the hot case.
Therefore, we created a Matlab routine that generates automatically the
heat flow map from ESATAN output file to synthesized the results.
Figure 6.16 summarized the results of the first hot case. Each of the
twelve components represented in this diagram involves several underly-
ing nodes so that temperatures and heat flows are averaged. All the exter-
nal parts of the CubeSat are at the top while the PCB stack is at the bottom,
encircled by the red dotted rectangle. The three red arrows represent the
total effective power exchanged radiatively between the PCBs and the ex-
ternal parts. The different radiative exchanges occurring between the PCBs
are not represented but can be estimated by differentiating the power in-
coming and leaving conductively one PCB being approximatively equal to
the power evacuated radiatively by the PCB.
Solar cells (32°C)
P
elec
= 3.56 W
12.52 W 11.99 W
Aluminum panels (32.1°C)
3.65 W 3.68 W
Aluminum frame (32.3°C)
3.17 W 3.76 W
Antenna deployment
mechanism panel (32.8°C)
0.15 W 0.43 W
Antennas (126.7°C)
0.61 W 0.23 W
3.03 W 0.36 W
2.89 W 0.11 W
OBC1 (36.3°C) COM (36.4°C)
EPS1 (55.1°C) OBC2 (39.5°C) EPS2 (39.3°C)
Batteries (48.1°C)
BAT PCB (48°C)
3200 mW
1740 mW
(face 3)
1460 mW
face 2 : 48%
face 5 : 52%
1630 mW
Spacers : 57 %
Bus : 43 %
1400 mW
Spacers : 75 %
Bus : 25 %
1180 mW
Spacers : 61 %
Bus : 39 %
1140 mW
Spacers : 29 %
Bus : 71 %
0 mW
40 mW
20 mW
170 mW 30 mW 170 mW
3.56 W
Legend
: Radiative
: Conductive
: Incident power
: Evacuated power : Dissipated power
Figure 6.16 – Hot case 1 heat ow map
This diagram leads to the following observations:
- The links joining the PCBs gives also the distribution of the heat flow
between the spacers and the connector. This highlights that both are
as important as the other.
- The solar cells are the most exchanging surfaces rejecting nearly as
much power as they receive (about 12 W).
Lionel Jacques
2
nd
Master in Aerospace Engineering
83 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
Figure 6.17 – Hot case 1: Hot spot due to the dissipation transistor and temperature
distribution on the antennas
- The mean electrical power collected is more than 3.5 W and is di-
rectly and entirely dissipated on the EPS.
- The global radiative exchange between the PCB stack and the struc-
ture is small compared to conductive one: about 10% (0.37 W /
3.22 W).
- Antennas are quite hot: their mean temperature is about 125

C but
the corresponding panel remains cold. 20 mW flow through the
coaxial cable to the amplifier. Figure 6.17 (right) details the temper-
ature distribution on the antennas and shows that the maximum
reached temperature occurs at the extremity: about 200

C.
- Hot spots are not visible on this diagram but the mean temperature
of the EPS PCB is 55

C while other PCB’s temperature ranges from
35 to 40

C. The left picture of Figure 6.17 shows that the temperature
near the transistor rises up to 115

C.
- The batteries are out of their allowed range of temperatures (in
charge): 48

C. It is worst than that was predicted by the STM, mainly
because the hot case is different (in the STM, the dissipated power
was simply divided between the COM and the EPS because a more
accurate distribution was not available)
- The conductive heat flux between EPS and BAT PCBs is null on av-
erage because the four spacers balance each other.
- The total conductive heat flux flowing through the bus and spacers
increases from EPS PCB to the outer PCBs because the radiative heat
flux between to consecutive PCBs is added at each step.
6.4.4 Hot Case 2
When analyzing the second hot case where the dissipated power is
less localized, one observes that OBC, OBC2 and EPS PCBs are cooler
on average while EPS2 and COM are hotter (Figure 6.18). Concerning the
structure, it has approximatively the same temperature as the first hot case
Lionel Jacques
2
nd
Master in Aerospace Engineering
84 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 6. Detailed Thermal Model
except for the antennas panel being a little hotter, like the antennas. This
is due to the amplifier and the coaxial cable, through which 300mW (of
the 1750mW dissipated by the transistor) flows. These are approximate
numbers because the coaxial cable is assumed to be directly connected to
the amplifier.
The batteries are cooler than in the first case but yet remain unaccept-
ably close to the 45

C allowed temperature in charge.
Solar cells (31.9°C)
P
elec
= 3.56 W
12.52 W 11.98 W
Aluminium panels (32°C)
3.65 W 3.68 W
Aluminium frame (32.3°C)
3.17 W 3.76 W
Antenna deployment
mechanism panel (33.4°C)
0.15 W 0.44 W
Antennas (130.3°C)
0.61 W 0.23 W
3.02 W 0.68 W
2.88 W 0.41 W
OBC1 (34.7°C) COM (44°C)
EPS1 (43°C) OBC2 (36.6°C) EPS2 (40.1°C)
Batteries (41.1°C)
BAT PCB (41.1°C)
2880 mW
1170 mW
(face 3)
1710 mW
face 2 : 53%
face 5 : 47%
1060 mW
Spacers : 55 %
Bus : 45 %
610 mW
Spacers : 121 %
Bus : −21 %
160 mW
Spacers : 165 %
Bus : −65 %
870 mW
Spacers : 31 %
Bus : 69 %
20 mW
20 mW
300 mW
170 mW 20 mW 170 mW
1.41 W 50 mW 300 mW
50 mW 1750 mW
Legend
: Radiative
: Conductive
: Incident power
: Evacuated power : Dissipated power
Figure 6.18 – Hot case 2 heat ow map
Little loops in the heat flow also appears, for instance between EPS1
and EPS2 where 160mW flows effectively from EPS1 to EPS2 but 264mW
through the spacers and 104mW through the connector in the opposite di-
rection. These local effects could not be revealed in the precedent thermal
model.
Figure 6.19 shows the local hot spots on EPS and COM PCBs. By de-
fault, the amplifier is located at the center of the PCB. With its efficiency
of only 30% involving a 1.75W dissipated power, the temperature of the
amplifier reaches 70

C. Once again, this temperature is relatively close to
the upper limit of the industrial temperatures range (+85

C) and must be
carefully looked at.
6.5 Parametric analysis and design
Previous results have confirmed two problems already spotted thanks
to the STM but revealed two others. The four issues for which solutions
have to be found are namely:
Lionel Jacques
2
nd
Master in Aerospace Engineering
85 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
Figure 6.19 – Hot case 2: Hot spots due to the dissipation transistor and amplifier on
COM PCB
1. In the cold case, the temperature of the batteries drops below its
acceptable lower limit at the end of the eclipse. An this is indeed the
more critical time: when they come into sunlight and get recharged,
involving the narrower range of temperatures (0 +45

C). A efficient
way of heating the batteries will thus be investigated in this section.
2. In the hot case, when maximal dissipation occurs, the temperature
of the batteries is too high. A way to enhance heat rejection must be
found to reduce the temperature reached by the batteries
3. The second issue concerns the dissipation transistor. According to its
datasheet, the main problem is not the transistor itself because it is
designed to support such temperatures but the components close to
it. Nevertheless, avoiding hot spots and reducing the temperature of
the EPS PCB would be favorable for the transistor as well as for the
batteries.
4. In the second hot case, the amplifier reaches too high temperature.
Nevertheless, as the amplifier is not yet determined and could be
much more efficient (like the ADL5541 linear amplifier which has
a constant 0.5 W power dissipation [4]), concrete measures should
not be taken before a better definition of the amplifier because they
could be inadequate if it was effectively more efficient. This problem
will thus not be plainly investigated in this work but should be as
soon as more information about the amplifier are available.
6.5.1 Cold case issues
The main problem of the cold case is the temperature of the batteries
dropping to -15

C when the satellite comes out of the eclipse, time at
which the batteries begin to charge again. This is thus the lower limit of
the charge temperature range that must be considered i.e. 0

C.
Lionel Jacques
2
nd
Master in Aerospace Engineering
86 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 6. Detailed Thermal Model
The use of a heater seems the easiest and more convenient way to
prevent the temperature from dropping below this limit. Nevertheless,
when designing a heating system, several questions arise:
How perform the control of the heater?
What is the required power?
Because the response is obviously not instantaneous, the batteries
will keep getting cooler a short time after the heater is switched on.
Then what is the optimal threshold below which it is turned on?
The power generated by the heater is not entirely accumulated in the
batteries: there are losses which does not participate to the heating
of the batteries. How to reduce these losses and thus increase the
heater efficiency?
Concerning the way to control the heater, two options are possible.
The first one involves the On-Board Computer (OBC) and allow to take
into account other criteria in addition to the temperature of the batter-
ies such as their voltage. The decision of whether it is wise to heat the
batteries when they are already nearly discharged is then more complex.
The second way of control consist in bypassing the OBC and develop an
independent and autonomous control loop. After many discussions with
the other subsystems and other CubeSat teams, we preferred the second
solution: a too cold battery is useless even if it has a low voltage and it
is more reliable in case of OBC failure (the heating system is still opera-
tional and the batteries can still feed the beacon). Figure 6.20 illustrates
the decreasing efficiency and capacity of a typical LiPo battery (here the
KOKAM SLPB723870H4) with the temperature. The -10

C and 0

C curves
illustrate than a heating system is well required.
Figure 6.20 – Battery temperature profile of a typical LiPo battery, here the KOKAM
SLPB723870H4 [32]
Moreover, EPS [55], EPS2 [33] designed the batteries in such a way that
their capacity should not drop below 70%, and this has been confirmed by
MIAS simulations.
Lionel Jacques
2
nd
Master in Aerospace Engineering
87 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Chapter 6. Detailed Thermal Model
Figure 6.21 shows the resulting control loop including the sensor and
the heater: if the temperature of the batteries drops under a given thresh-
old, the heater is switched on until the batteries are again within its allow-
able temperature range.


No
Yes
Heater off
Temperature sensor
>

?
Heater on
Figure 6.21 – Heating system control ow chart
As the batteries are located on either side of the BAT PCB, we decided
to use two separated redundant heaters in parallel with two sensors to
increase the reliability of the system. Actually, a third sensor should be
used: if two sensors give different temperatures, the third one enables
the defective sensor determination. On another hand, using two identical
heaters in parallel ensures that if one fails, the other is still operational.
To answer the three other questions about the required power, thresh-
old and losses reduction, parametric analyses are performed. Diminishing
the losses is achieved by insulating the batteries and their PCB. As early
presented in this work, only two ways of heat transfer occur in space:
radiation and conduction (except when considering aerothermal flux or
specific thermal control means such as fluid loops or phase-change ma-
terials). From this observation, two corresponding ways of insulating the
batteries have been foreseen and investigated:
covering the batteries with an aluminum tape (low emissivity) to
reduce radiative losses,
and using insulating materials for the fasteners to reduce conduction
losses.
Aluminum tape is easy to use and largely used in space applications.
According the the Sheldahl Red Book p.33 [52], Sheldahl’s first surface
aluminized polyimide tape with 966 acrylic adhesive has a low emissivity
( 0.03) is low outgassing and would be suitable.
Concerning the BAT PCB fasteners, usual insulating materials are ther-
moplastics such as PTFE (Teflon
R
) or PA (polyamide such as nylon) but
these have poor structural properties and outgassing must be carefully
looked at. Fully replacing the initial aluminum spacers by a thermoplastic
materials must then be avoided because the BAT PCB undergoes strong
dynamic amplifications (being in the center of the CubeSat, fastened to an-
other PCB). Furthermore, structural integrity could be weakened because
of the creep phenomenon. For those reasons, in agreement with STRU sub-
system, here is the outcoming design depicted in Figure 6.22. Structural
integrity is enforced by titanium spacers and screws (high tensile strength
Lionel Jacques
2
nd
Master in Aerospace Engineering
88 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
and Young’s Modulus to thermal conductivity ratio) in addition to thin
nylon washers, reducing creep issues and increasing thermal insulation.
Nylon has been preferred because of its higher Young’s Modulus.
washers
BAT PCB
EPS PCB
Titanium screw
Titanium spacer

Figure 6.22 – Batteries’ PCB fastening concept
The resulting insulation gain can easily be evaluated in considering
only the main heat transfer through the spacer (conduction through the
screw is neglected because of poor contact around the nut) and a perfect
contact between the PCBs and washers. The equivalent resistance com-
pared to an aluminum spacer of the same dimensions is far much greater:
R
alu
=
L
spacer
k
alu
A
spacer
& R
eq
=
_
L
spacer
2 t
washer
k
titanium
+
2 t
washer
k
washer
_
1
A
spacer
=
L
spacer
k
eq
A
spacer
This defines the equivalent thermal conductivity which will be used as
parameter:
k
eq
=
k
washer
k
titanium
k
washer
(1 γ) + k
titanium
γ
where γ =
2 t
washer
L
spacer
. According to CES software, the thermal conductiv-
ity of PA (nylon 6.6) is 0.24 W/mK. Based on this value, the Figure 6.23
shows the influence of the washer thickness and thermal conductivity of
the spacer. The left plot demonstrates that 0.5 to 1 mm thick washers al-
ready reduce efficiently the conduction and using thicker washers would
be useless. The right plot shows that for thick washers (above 1 mm), the
thermal conductivity of the spacer is less influent than for thinner wash-
ers. For instance, considering 0.5 mm thick washers replacing aluminum
(k = 150 W/mK) by titanium (k = 10 W/mK, CES Edupack value) reduce
the equivalent thermal conductivity from 3.5 to 2.7 W/mK, i.e. a 23% gain.
Eventually, considering 0.5 mm thick PA washer combined with
titanium spacer, the global thermal conductance is now equal to
1.32 10
3
W/K while the initial one, with the aluminum spacer, was
73.5 10
3
W/K. The conductance is thus divided by a 55 factor.
Now that solution have been proposed, their concrete effects are stud-
ied. After implementing the heating system in ESATAN, the insulation
benefit is quantified. The threshold is set to 5

C and the heater power is
varied (the total heater power for the two batteries). Figure 6.24 shows the
minimum temperature reached on the batteries in function of the heater
power for different insulating configuration. The effect of radiative in-
sulation is clearly observed: for a same equivalent thermal conductivity
Lionel Jacques
2
nd
Master in Aerospace Engineering
89 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
50
100
150
1
2
3
4
5
2
4
6
8
10
12
14
16
washer thikness [mm]
spacer thermal conductivity [W/mK]
e
q
u
i
v
a
l
e
n
t

t
h
e
r
m
a
l

c
o
n
d
u
c
t
i
v
i
t
y

[
W
/
m
K
]
0 50 100 150
0
2
4
6
8
10
12
14
16
18
20
spacer thermal conductivity [W/mK]
e
q
u
i
v
a
l
e
n
t

t
h
e
r
m
a
l

c
o
n
d
u
c
t
i
v
i
t
y

[
W
/
m
K
]


0.1 mm washer
0.2 mm washer
0.5 mm washer
1 mm washer
2 mm washer
5 mm washer
Figure 6.23 – Equivalent conductivity of BAT PCB fastening in function of the PA
washer thickness and thermal conductivity of the spacer
100 200 300 400 500 600 700 800 900
−15
−10
−5
0
5
10
Heater Power [mW]
B
a
t
t
e
r
y

M
i
n
i
m
a
l

T
e
m
p
e
r
a
t
u
r
e

[
°
C
]
,

T
h
r
e
s
h
o
l
d

=

5
°
C


high ε
battery
, k=2
low ε
battery
, k=2
low ε
battery
, k=10
low ε
battery
, k=100
Figure 6.24 – Effect of insulating the batteries on the heater required power for a given
threshold [5

C]
(2 W/mK) the blue doted line corresponding to the high emissivity (the
initial value considered for the batteries was 0.8 i.e. plastic) shows that
a more than 500 mW heater is required to reach the threshold. On the
other hand, reducing the emissivity of the batteries through aluminum
Lionel Jacques
2
nd
Master in Aerospace Engineering
90 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
tape ( = 0.03) lowers the required power to 300 mW. Not reducing the
emissivity of the batteries is nearly equivalent to increase the equivalent
thermal conductivity to 10 W/mK as shows the red curve. Finally, the
black curve demonstrate that even a 1W heater would not be sufficient
if the conductive insulation is not achieved. Figure 6.24 also shows that
increasing the insulation of the batteries increases the heater sensibility
(
dT
batteries
dP
heater
): the slope of the 2 W/mK curve (blue) is greater than the slope
of the 10 W/mK or 100 W/mK one (before the threshold is reached).
Two of the four initial questions have been answered. The remaining
two questions concern the required power and the switching threshold.
Three parameters are looked at in this study:
- the minimal temperature reached by the batteries to verify that the
power is sufficient,
- the time during which the heater is turned on,
- and the consumption of the heating system.
The consumption of the heating system is computed as follows: the
energy of the battery is given by its capacity, given in mAh. The power
generated by the heater is thus multiplied by the time during which it is
turned on and then divided by the voltage.
C =
P t
U
[mAh]
The heater will be directly connected on the batteries to enhance relia-
bility. The voltage is therefore not constant because it is not converted by
one of the EPS converter. Figure 6.20 showed that the maximum voltage is
4.2 V while cut-off voltage is 2.7 V. The worst case corresponds therefore
to the minimum voltage: a 2.5 V has been assumed to be conservative.
Figure 6.25 shows the evolution of the three parameters presented
above in function of the power and for different thresholds. Top plot con-
firms that the lower the threshold, the low the require power to reach it.
Bottom plot shows that reducing the threshold and increasing the power
reduces the time the heater is turned on. For a 500 mW heater, the time
it is turned on can be reduced to 35 to 50 min, depending on the thresh-
old, i.e. 35 to 50% of the orbital period (104 min) while the maximum
eclipse time is about 35 % of the orbital period. The middle plot shows
that the increase power resulting in a reduced consumption time balance
each other once the threshold is reached: the capacity curves are approxi-
matively horizontal (due to computation error of the consumption time
1
)
after that point. Middle plot also shows that reducing the threshold from
5 to 0

C allows to save 40 mAh.
Nevertheless, the thermal sensors and heater have a finite accuracy
and LiPo battery’s characteristic curves illustrated in Figure 6.20 suggest
therefore that sufficient safety margins to cope with these uncertainties
combined to the model and environmental uncertainties. The 40 mAh
consumption that could be saved by reducing the threshold remains quite
1
The time the heater is on is obtained through the post-processing of the ESATAN
output file. The time the heater is turned on consequently depends on the output time
step, set here to 360 points per orbit
Lionel Jacques
2
nd
Master in Aerospace Engineering
91 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
small compared to the nominal 1000 to 1500 mAh capacity of the foreseen
batteries, recalling that the worst case has been assumed with the 2.5 V
voltage.
100 150 200 250 300 350 400 450 500
−5
0
5
B
a
t
t
e
r
y

M
i
n
i
m
a
l

T
e
m
p
e
r
a
t
u
r
e

[
°
C
]


Threshold = 0°C
Threshold = 1°C
Threshold = 2°C
Threshold = 3°C
Threshold = 4°C
Threshold = 5°C
100 150 200 250 300 350 400 450 500
60
80
100
120
140
160
C
a
p
a
c
i
t
y

[
m
A
h
]
,

U
=
2
.
5
V
100 150 200 250 300 350 400 450 500
20
40
60
80
100
120
Power [mW], low ε
battery
, k
BAT
=1W/mK
T
i
m
e

t
h
e

h
e
a
t
e
r

i
s

o
n

[
m
i
n
]
Figure 6.25 – Evolution of the battery minimal temperature, batteries’ capacity decrease
(worst voltage U=2.5 V) and time the heater is turned on in function of the heater power
for different thresholds
For the sake of reliability, the threshold shall be set to 5

C and we shall
use a 500 mW heater. For the same reason, instead of using one heater
for the two batteries, each battery shall have its own heater as previously
stated. The 500 mW have to be divided into two 250 mW heaters. Figure
6.25’s top plot shows that even if one fails, the remaining 250 mW are able
to ensure that the batteries’ temperature will not go below 0

C.
Moreover, vacuum tests have been conducted by EPS subsystem on dif-
ferent batteries (KOKAM and Varta) to verify their functionality in their
allowable temperature range. Results are available in the test report [34].
A persistent deformation has been noticed on both models so that batter-
ies should be packed. In addition to that, even if this was not the subject
of these tests, heating has also been noticed under discharge. Figure 6.26
illustrates increase of temperature: about 5

C whereas the batteries are
fastened to a temperature controlled panel through a nitrogen loop. The
same phenomenon has been observed for the VARTA PoLiFlex
R
battery.
However, the determination of the generated power is impossible through
this data because this was not the purpose of this test. Nevertheless, the
conclusion is that the batteries undergo a self heating during discharge
Lionel Jacques
2
nd
Master in Aerospace Engineering
92 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
which is favorable: when supplying the heaters, additional heat is auto-
matically generated inside the batteries.

Temperature decrease
before discharge
End of discharge and
temperature decrease
5.2°C temperature increase
because of discharge
Beginning of
discharge
T
°

[
K
]

Time
Figure 6.26 – Temperature of a SLB 603870H KOKAM battery during the cold test [34]
To determine the resistance of the heaters, the lowest voltage is once
again assumed. Using Ohm’s law, the electrical resistance is equal to
R =
U
2
P
=
2.5
2
0.25
= 25 [Ω]
As the battery model is still unknown, it is difficult to determine
the heater model. However, based on the typical dimensions of the dif-
ferent foreseen models, we found the following MINCO reference [40]:
XHK5377R26.3L12B where
- X means low outgassing ink
- HK stands for polyimide heater
- 5377 specifies for the model size and shape: 35.6 59.4 mm rectan-
gular shape
- R26.3 specifies the resistance value : 26.3 ω (This was the closest
value from the one defined above corresponding to the dimensions
of the batteries)
- L12 is the lead length in inches (12 inches is the standard length,
other length are possible on demand).
- B specifies Pressure-sensitive adhesive (PSA)
If it appears that the batteries should be packed in a thin alu-
minum box with epoxy resin, adhesive could possibly be removed
(XHK5377R26.3L12A).
Lionel Jacques
2
nd
Master in Aerospace Engineering
93 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
6.5.2 Hot case issues
In order to deal with the hot case issues, three solutions have been
investigated:
1. In agreement with the EPS subsystem, we suggested to relocate
a part of the power initially dissipated through the transistor by
adding a resistance in series with the transistor and fastening it
onto the antenna deployment mechanism panel. For this purpose,
the use of an adhesive Minco Thermofoil
TM
heater [40] seamed con-
venient thanks to their easy integration and wide range of shapes.
The EPS subsystem determined that the ideal resistance to prevent
the transistor from depolarizing was 2.4 Ohms. Because this resis-
tance value was not available among suitable shapes, we decided to
use two 4.7 Ω resistances in parallel, involving a equivalent resis-
tance of 2.35 Ω. With this configuration and according to the EPS
subsystem, the power distribution between the equivalent resistance
and the transistor is depicted in Figure 6.27. The bottom plot shows
that the power dissipated by transistor is already divided by two for
the 3.5 W mean collected power computed and shown in Figures
6.18 and 6.16.
0 1 2 3 4 5 6
0
1
2
3
4
5
Total dissipated power [W]
P
o
w
e
r

[
W
]


Resistances
Transistor
0 1 2 3 4 5 6
0
20
40
60
80
100
Total dissipated power [W]
P
e
r
c
e
n
t
a
g
e

o
f

t
o
t
a
l

d
i
s
s
i
p
a
t
e
d

p
o
w
e
r

[
%
]
Figure 6.27 – Power distribution in the equivalent resistance and the transistor of the
dissipation system
2. To reduce the temperature of the batteries, the second possible so-
lution investigated, as already introduced, is to fasten the batteries’
PCB to the EPS2 PCB instead of the EPS one because the EPS2 PCB is
less dissipative and cooler. The effect of insulating batteries’ PCB, as
introduced above for the cold case issues, is also investigated. Two
equivalent thermal conductivities are considered: 150 W/mK (full
aluminum spacers) and 2 W/mK (combination of titanium screws
Lionel Jacques
2
nd
Master in Aerospace Engineering
94 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
and spacers with Nylon or PTFE washers). As radiative insulation
has proved to be useful in the cold case, low emissivity is assumed
(batteries covered with aluminum tape).
3. The third solution is to use a thermal strap between the transistor
and the antenna deployment mechanism panel, in agreement with
STRU and MECH subsystems. We designed a copper angle bracket
that would be strongly bolted on the EPS PCB with two M3 bolts,
encircling the transistor, then located close to the face 4. Figure 6.28
shows the angle bracket in red encircling the transistor (left) on EPS
PCB engineering model and a CATIA model illustrating the two ad-
hesive resistances and angle bracket configuration.

Resistances
in parallel
Thermal strap
(Angle bracket)
Figure 6.28 – Thermal strap (angle bracket in red) design on engineering EPS PCB
(left) and its fastening configuration with the relocated resistances on antenna panel
(CATIA Model on the right)
The effects of these three solutions are summarized in Table 6.3 for the
different combinations. A combination is described through the first four
columns in which are given the PCB to which the BAT PCB is fastened
(EPS/EPS2) and the fastening equivalent thermal conductivity (k
BAT
) the
strap column shows whether the transistor is strapped to the antennas’
panel or not. The different combinations are compared through the bat-
teries and transistor temperatures. The effect of relocating a part of the
power through the two Minco resistances is given in the last two columns.
The first row recalls the initial design (first two columns) and shows
the effect of the relocated power without any other change. The transistor
temperature is reduced by a 1.5 factor while a 7

C battery temperature
decrease is already achieved.
The next eight rows assumes that the batteries are covered with alu-
minum tape as it has proved to be useful when examining the cold case.
Among these eight rows, the first half gives the temperature for the differ-
ent combinations without strap. The effect of the equivalent thermal con-
ductivity starts to be significant when the BAT PCB is fastened to EPS2
Lionel Jacques
2
nd
Master in Aerospace Engineering
95 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
and a higher thermal conductivity would be preferred if it should not
be avoided for the reasons exposed in the cold case analysis. This can be
easily explained: the heat radiatively received by the batteries must be
evacuated to reduce their temperature. A slight increase of the transistor
temperature is also noticed.
The second half of the table shows the results when the thermal strap
is added and demonstrates that the combination of the strap with the
Minco resistances can reduce the temperature of the batteries to 35

C and
the temperature of the transistor to 67

C, i.e. less than 60% of its initial
value. Even if fastening the batteries’ PCB to EPS2 does not appear to be
anymore worthwhile when strapping the transistor, reliability and uncer-
tainties suggest to combine both solutions.
without resistances with resistances
strap EPS/EPS2 k
BAT
T
batteries
T
transistor
T
batteries
T
transistor
[W/mK] [

C] [

C] [

C] [

C]
high
bat
no EPS 150 48 114 41 75
low
bat
no EPS 150 48 115 41 76
no EPS 2 47 115 40 77
no EPS2 150 39 115 36 77
no EPS2 2 45 115 39 77
yes EPS 150 38 94 37 67
yes EPS 2 37 95 35 67
yes EPS2 150 35 95 34 67
yes EPS2 2 36 95 35 67
high aluminum panel (0.8)
high
bat
no EPS 150 46 112 43 80
low
bat
no EPS 150 47 113 43 80
no EPS 150 35 94 34 66
Table 6.3 – Effects of the different solutions to the hot case issues
To make the temperature more uniform inside spacecraft, black paint
is usually used on internal structure elements to increase their emissivity
(being low because spacecraft’s structure is usually made of aluminum).
On OUFTI-1, only the aluminum panels can be coated with a black paint
because Pumpkin Kit’s Structure is not intended to be modified. The emis-
sivity of the 6 aluminum panels has therefore been set to 0.8 (inside) and
the results are available in the last three row of Table 6.3 for three com-
binations: the nominal case, the nominal case with batteries covered with
aluminum tape and the final design.
High emissivity on both aluminum panels and batteries allow a 2

C
decrease of the batteries and transistor temperatures in the nominal design
and a 1

C decrease if the batteries are covered with aluminum tape.
In the last combination, black paint enables another 1

C decrease. A
black coating of the aluminum panels implies therefore not enough signif-
icant changes in the temperature to be worthwhile.
Finally, Figure 6.29 shows the heat flow map resulting from the modifi-
cations discussed above i.e. strapping the transistor to the antennas’ panel,
relocating of a part of the dissipation system power and insulating the
Lionel Jacques
2
nd
Master in Aerospace Engineering
96 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
batteries with aluminum tape. It shows that nearly one half of the power
is relocated and that 1 W of the 1.78 W remaining dissipated power on
the transistor is directly evacuated through the strap and less than 1 W
has still to be evacuated through the spacers. The power incident to the
amplifier is however slightly increased because the antennas’ panel is hot-
ter (80 mW instead of 20 mW)) but this has no significant effect on the
temperature of the amplifier. Table 6.4 summarized the achieved tempera-
tures. We also verified that EPS2 fastening has no significant effect on the
heating system.
Solar cells (31.8°C)
P
elec
= 3.56 W
12.52 W 11.95 W
Aluminum panels (31.9°C)
3.65 W 3.67 W
Aluminum frame (32.6°C)
3.17 W 3.78 W
Antenna deployment
mechanism panel (37.5°C)
0.15 W 0.46 W
Antennas (129.2°C)
0.61 W 0.23 W
2.99 W 0.3 W
2.97 W 2.75 W
OBC1 (33.2°C) COM (33.9°C)
EPS1 (36.7°C) OBC2 (34.2°C) EPS2 (34.2°C)
Batteries (34.9°C)
BAT PCB (34.9°C)
790 mW
440 mW
(face 3)
350 mW
face 2 : 59%
face 5 : 41%
400 mW
Spacers : 73 %
Bus : 27 %
280 mW
Spacers : 91 %
Bus : 9 %
250 mW
Spacers : 89 %
Bus : 11 %
360 mW
Spacers : 75 %
Bus : 25 %
0 mW
0 mW
80 mW
40 mW 10 mW 50 mW
1.69 W
1.87 W
1090 mW
Legend
: Radiative
: Conductive
: Incident power
: Evacuated power : Dissipated power
Figure 6.29 – New design hot case heat ow map
Concerning the COM amplifier, because too high uncertainties remains
about the dissipated power, no design modifications were investigated.
We only studied the effect of the location of the amplifier on the PCB and
its dissipated power. To avoid numerous costly ESATAN simulations, the
parametric analysis has been carried out through some simplifications.
Radiation is neglected and the PCB is assumed to be an insulated plate
subjected to some boundary conditions and internal heat loads. Using the
finite difference method, this kind of problems can be reduced to a single
matrix equation:
A T = g
where T is the 1 n vector containing the n temperatures of the discretized
problem, g the heat loads incident on each node and A is a n n matrix
represents the conductive coupling between the nodes. For the sake of
clarity, the underlying equations are not detailed here but are available in
appendix E.
Lionel Jacques
2
nd
Master in Aerospace Engineering
97 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
Element T

before [

C] T

after [

C]
Solar cells 32 32
Aluminum panels 32 32
Aluminum frame 32 33
Antennas’ panel 33 38
Antennas 127 129
OBC PCB 36 33
OBC2 PCB 40 34
EPS PCB 55 37
EPS2 PCB 39 34
COM PCB 36 34
Batteries 48 35
Dissipation Transistor 114 67
Table 6.4 – Comparison of hot case temperatures before and after modifications
0
0.2
0.4
0.6
0.8
1
0
0.5
1
0
10
20
30
40
50
60
70
relative position of the hot spot in x direction
re
la
tiv
e
p
o
s
itio
n
o
f th
e
h
o
t s
p
o
t in
y
d
ire
c
tio
n
r
e
l
a
t
i
v
e

t
e
m
p
e
r
a
t
u
r
e

i
n
c
r
e
a
s
e

[
°
C
]
Figure 6.30 – Amplifier temperature increase in function of its relative location on the
COM PCB
Based on this model, the COM PCB has been discretized and the tem-
peratures of its four corners are supposed to be constant which is not far
from reality. The effect of local heat dissipation has been studied in func-
tion of its location on the PCB. Figure 6.30 shows the temperature increase
on the amplifier in function of its relative position on the PCB. The dis-
sipated power is maintained at 1.75 W. The worst locations involving the
greater increase of temperature are the center of the four edges of the PCB.
But getting closer to the corners reduces the temperature of the amplifier.
It is worthwhile to notice that setting the amplifier at the center of the PCB
is not the worst situation (50

C temperature increase in comparison to the
Lionel Jacques
2
nd
Master in Aerospace Engineering
98 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
corner’s temperature) and that this corresponds indeed to the temperature
observed in ESATAN results (Figure 6.19), 75

C while the temperature of
the corners is about 30

C, i.e. a 45

C increase. One must bear in mind
that the bus has been neglected as well as the radiative exchanges, which
explains the difference.
The use of another amplifier such as the ADL5541 linear amplifier
which dissipates a constant 0.5 W power [4] has also been investigated.
The resulting temperature increase drops down to only 15

C at the center
of the PCB (and the reached temperature given by ESATAN is about 45

C).
Summary
The Detailed Thermal Model, far much complicated and computation-
time costly than the previous one, has however proved worthwhile. The
issues already revealed by the Simplified Thermal Model became more
tangible and evident. On another hand, the present model enable new
issues detection, especially in the hot case. The four issues that have been
confirmed or revealed are namely:
- the too cold batteries in the cold case;
- their too high temperature in the hot case;
- the hot spot due to the dissipation transistor;
- the too high temperature reached on the amplifier.
Concrete measures have been taken to solve three of these four issues.
Only three because too many uncertainties still surround the COM am-
plifier. However, we stated that the efficiency of the amplifier should be
taken into account when selecting the amplifier. For instance, using the
ADL5541 could reduce the temperature down to 45

C.
A heating system has been designed to cope with batteries’ too cold
temperature. Two 250 mW heaters in parallel, one for each battery, are re-
quired to maintain the temperature above the threshold, which shall be set
to 5

C because of sensor and comparator finite accuracy. Bearing in mind
the problems encountered by Compass-1, we suggest to use three tempera-
ture sensors for reliability purposes. Insulating the batteries has proved to
be required to reduce losses and consequently increase the heating system
efficiency. The foreseen Minco heaters has not yet been ordered because
the definitive battery model is not defined. However, even if the batteries
are not defined, comprehensive tests should be conducted to validate the
design a evaluate the sensors and comparator accuracy.
Concerning the hot case, the problems were involved by the dissipation
system. Two main measures have been investigated to reduce both batter-
ies and dissipation transistor temperatures: an angle bracket encircling
the transistor and fastened to antennas’ deployment mechanism panel al-
ready allows a 10

C temperature decrease of the batteries (from 48

C to
38

C) and 20

C of the transistor (from 115

C to 95

C). In addition to that,
a part of the dissipated power is now relocated through two resistances
in parallel also fastened to the antennas’ panel. This involves another 3

C
and 30

C temperature decrease of the batteries and transistor, respectively
Lionel Jacques
2
nd
Master in Aerospace Engineering
99 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 6. Detailed Thermal Model
reduced to 35

C and 67

C. Once again, tests should be conducted to eval-
uate the real efficiency of these measures.
Lionel Jacques
2
nd
Master in Aerospace Engineering
100 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
7
Testing
In the previous chapters, the different models revealed that the temper-
atures of some components go beyond their limits. Thanks to the Detailed
Thermal Model, proper measures have been taken to ensure those compo-
nents remain within their temperature limits. However, unlike electronic
equipments that can be ground tested, the thermal behavior of a satellite is
very difficult to verify on Earth. This behavior can thus only be predicted
by thermal analytical methods and can only be fully verify once it is in
orbit.
Consequently, thermal modeling involves many assumptions and un-
certainties and the design relies on a worst case approach. Tests are thus
essential for the verification and to ensure the satellite achieves all the
requirements.
This chapter intends to describe the tests that shall be performed at the
Centre Spatial de Liège (CSL). First, the different test and model philoso-
phies are presented. The chosen one is then detailed and the correspond-
ing temperature ranges are defined. Next, the typical test hardware is ex-
plained along with the foreseen set-up.
7.1 Test and model philosophy
For several decades, many documents have been written to standardize
definitions, tests requirements and methods, especially for military space
vehicles. Those documents also introduced test categorization and levels.
Today, three test categories are established:
1. Engineering or Development Tests are conducted throughout a satel-
lite design in each subsystem to validate new design concepts or
perform measurements to reduce uncertainties.
2. Qualification Tests are conducted to demonstrate that the design im-
plementation and manufacturing process have resulted in hardware
and software that meets specification requirements with sufficient
margins.
3. Flight Acceptance (FA) Tests are conducted to verify conformance
to specification requirements and provides quality-control assurance
to detect workmanship deficiencies, manufacturing errors or any la-
tent defect that would be detected by by normal inspection tech-
niques. Acceptance tesing is less severe than qualification testing and
101
Chapter 7. Testing
are conducted under environmental conditions no more severe than
those expected during the mission.
It is important to note that items which have undergone qualification
testing are usually not eligible for flight because the remaining life regard-
ing fatigue and wear is not demonstrated. However, tight schedules, bud-
getary constraints are such as a dedicated nonflight qualification model is
not always feasible. For this reason, alternative test strategies and trade-
offs have been developed to reduce the risks involved by this situation.
These alternative strategies introduce higher risks compared to the nomi-
nal test program constituted of the acceptance test following the qualifica-
tion one but increasing safety factors and development tests can mitigate
the induced risks.
The main alternative strategy is protoqualification (or protoflight) test-
ing. The tested protoflight model (PFM) is thus considered eligible for
flight. Protoflight testing accomplishes thus in one test the combined pur-
poses of design qualification and flight acceptance and results from a
combination of the two strategies. Protoflight eliminates the redundancy
of building a qualification hardware, enables significant cost saving and
is thus well suited for student projects. This is the adopted strategy for
OUFTI-1. However, a second Pumpkin’s CubeSat Kit have been bought
for the engineering tests and model. It is therefore on the second CubeSat
Kit that measurements of chapter 5 were performed.
The above strategies concerned both thermal and vibration tests. More
specifically, thermal test are divided into three tests:
1. a Thermal Vacuum Test (TV) is performed under vacuum and sub-
jects the satellite (or equipment) to worst hot/cold temperatures in-
cluding adequate margins. Its purpose is the performance verifica-
tion through functional testing because it is the most realistic ground
simulation of the in-orbit environment.
2. a Thermal Cycling Test (TC), that can be performed under ambient
pressure, subjects the satellite (or equipment) to a series of cycles
of hot and cold temperature plateaus. Its main purpose is to reveal
latent workmanship defects due to environmental stress.
3. a Thermal Balance Test (TB) is generally conducted as a part of
the thermal vacuum test. It has two primary purposes: demonstrate
the ability of the thermal control system to maintain temperatures
within the specified operational limits and provide data for the TMM
correlation.
These three tests can be combined in one test called thermal vacuum
cycling test during which the TB test is also performed.
7.2 Tests specifications
Now that the tests have been introduced and before describing them
in details, temperature ranges and margins terminology is presented. As
previously introduced, the thermal model relies on several assumptions
and uncertainties. They can be categorized in three main classifications:
Lionel Jacques
2
nd
Master in Aerospace Engineering
102 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
1. Environmental uncertainties: albedo, Earth IR and direct solar fluxes
depending for instance on solar activity, clouds coverage. . .
2. Spacecraft physical uncertainties: thermo-optical and thermal prop-
erties of the materials, interface conductances. . .
3. Mathematical models (both GMM and TMM) uncertainties: the tem-
peratures of the spacecraft are governed by a set of nonlinear dif-
ferential equations for which no closed integration solution is avail-
able and numerical solutions are employed based on finite difference
methods. These methods have intrinsic uncertainties:
- The definition of finite, isothermal elements is physically incor-
rect.
- Conductive and radiative exchange between these isothermal
elements are only approximated (Monte-Carlo Ray Tracing,
thermal resistance across different section. . . ).
- Numerical integration of the differential equations system is
subjected to tolerances and consequently has only a finite accu-
racy.
To account for these uncertainties, safety margins are applied to the
worst case predicted temperatures. The resulting temperature forms the
basis for the acceptance temperature range. To this acceptance tempera-
ture, extra margin are added for protoqualification testing. Qualification
temperature range are usually the same as the protoqualification one but
additional qualification margins can be added to increase environmental
conditions over that expected during the lifetime. Qualifications margins
may also include the cycles duration as well as any other increase in sever-
ity to demonstrate the robustness of the design.
Figures 7.1 and 7.2 illustrates the different temperature range and
margin terminology for European and NASA/JPL programs respectively.
They are quite identical in the fact that acceptance range is obtained by
adding thermal control uncertainty margins (thermal design margins) and
residual margins (FA thermal reliability margin in NASA/JPL). The qual-
ification range accounts for a 10

C to 15

C additional margin.
Now that the qualification temperature range is clearly defined, two
other temperatures are used in the following figures and need to be de-
fined: the non-operating temperature (max and min) which are the high-
est/lowest temperature for an equipment to survive unpowered and the
start-up temperature which is highest/lowest temperature of the equip-
ment, at which it can be switched on.
Lionel Jacques
2
nd
Master in Aerospace Engineering
103 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing









FA thermal reliability margin (5°C)
Thermal reliability margin (15°C)
Thermal reliability margin (20°C)
FA thermal reliability margin (5°C)
Worst case
hot/cold
predicted
temperature
range
Thermal design margin
Thermal design margin
Allowable
flight
temperature
range
Flight
acceptance
temperature
range
Protoflight/
qualification
temperature
range
D
e
s
i
g
n

&

a
n
a
l
y
s
i
s

T
e
s
t
i
n
g

Figure 7.1 – Thermal margin terminology for NASA/JPL programs [24]
ECSS
15 February 2002
ECSS--E--10--03A
138
B.2.3 Institute of Environmental Sciences
Environmental Stress Screening Guidelines
ESSEH 1981.
B.2.4 USAF
MIL--STD--1540 B
Military Standard: Test Requirements for Space Vehicles.
B.3 Influence of equipment temperature limits on thermal design
The general definition of temperature limits and margins of a spacecraft thermal
design are summarized in Figure B--1. The bandwidth between the upper and
lower temperature limits defines the thermal designrange and the type of thermal
design to be used. The smaller the bandwidth, the greater the analytical effort is
performed to ensure the limits, and the more likely the requirement for an active
or semi-active control.
The approach in defining limits in certain cases is very different between US and
European programmes, with European programmes usually having the more
restrictive temperature limits. Further complications arise when a programme
uses certain equipment from a previous project that cannot exactly complement
the limits pertaining to the other equipment.
Ideally, a trade-off is made early in each project, either to
a. allowlarge temperature limit bandwidths, with the resulting costs and effort
to qualify the components, in conjunction with a simple passive thermal
control subsystem, or
b. to impose a narrow temperature limit bandwidth and transfer the costs and
the effort to the thermal control subsystemto provide anactive control system
to maintain equipment within limits.
Environmental design margin 10 ºC
Residual margin
Predicted temperature
range during service life
Thermal control uncertainty 10 ºC or 15 ºC
Thermal control uncertainty 10 ºC or 15 ºC
Residual margin
Environmental design margin 10 ºC
T
h
e
r
m
a
l
d
e
s
i
g
n
r
a
n
g
e
E
q
u
i
p
m
e
n
t
d
e
s
i
g
n
e
n
v
i
r
o
n
m
e
n
t
(2)
(2)
(3)
(3)
(1)
(1)
Maximum qualification test temperature
Minimum qualification test temperature
Upper temperature limit (max. flight acceptance test temp.)
Maximum expected temperature
Upper predicted temperature (worst case)
Minimum expected temperature
Lower predicted temperature (worst case)
Lower temperature limit (min. flight acceptance temp.)
Notes
b Thermal control uncertainty depends on type of equipment:
10 ºC uncertainty for thermal control verified by test,
15 ºC uncertainty for thermal control unverified.
3. Residual thermal design margin can be zero.
4 10 ºC margin does not include test condition tolerances
Figure B--1: Temperature limits and margins definitions
Figure 7.2 – Thermal margin terminology from ECSS-E-10-03A standard [17]
Lionel Jacques
2
nd
Master in Aerospace Engineering
104 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
ECSS
15 February 2002
ECSS--E--10--03A
74
T
NO-max
T
SU-high
T
Q-max
T
AMBIENT
T
Q-min
T
SU-low
T
NO-min
1 Cycle
t
E
time (hours)
T (ºC)
Mode
2
Mode
1
Mode
3
Mode
1
Mode
3
START-UP
P (hPa)
time (hours)
1 × 10
-4
1 × 10
-5
COMBINED CYCLING AND VACUUM TEST
AMBIENT
t
E
t
E
t
E
t
E
t
E
START-UP
T
AMBIENT
n Cycles
For explanation of symbols see Table 14
Pressure
Temperature
Figure 10: Equipment thermal vacuum cycling test sequence
Figure 7.3 – Thermal vacuum cycling test sequence [17]
Figure 7.3 illustrates a typical thermal vacuum cycling test sequence.
The test begins and ends with a functional test at ambient temperature.
Then, at the pressure of of 10
4
hPa, the test always starts with a temper-
ature increase up to the high non-operating level (T
NO,max
) to accelerate
outgassing (baking sequence). After a dwell time t
E
, the temperature is
decreased to the maximum start-up level (T
SU,high
) and is then stabilized
at the high operating temperature (T
Q,max
). Once the temperature is stabi-
lized (time t
E
) the functional and performance tests (including TB) are per-
formed and after that it is switched off again to decrease and maintained
at the minimum non-operating temperature (T
NO,min
) during a time t
E
.
The temperature is then increased to the minimum start-up temperature
and the equipment is switched on. When stabilized at the low operating
level (TQ
min
), and after the time t
E
, the functional tests in cold case are
performed.
The spacecraft is then cycled between T
Q,max
and T
Q,min
until the num-
ber of cycles specified in Table 7.3 is achieved. Functionally tests are again
performed during the last cycle at T
Q,max
and T
Q,min
after which the tem-
perature is raised to ambient conditions and the final functional and per-
formance test can be performed. All the symbols are summurized in Table
7.1
As defined above, the protoflight approach combines acceptance and
qualification: levels are the qualification ones while duration are as accep-
tance ones [17]. Nevertheless 6 cycles could be performed.
Lionel Jacques
2
nd
Master in Aerospace Engineering
105 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
ECSS
15 February 2002
ECSS--E--10--03A
66
h. When stabilized at the lowoperating level (T
Q--min
), and after the time t
E
, the
functional and performance test shall be performed.
i. The temperature and pressure shall be raised to ambient conditions
(T
AMBIENT
) and the final functional test performed.
NOTE Characteristic parameters of thermal vacuumtest are speci-
fied in table 15.
j. The temperature rate of change < 20 ºC/min shall apply only to equipment
within the space vehicle. For equipment outside the space vehicle, higher
gradients are specified in the appropriate equipment specifications.
Table 14: Legend and symbols for Figures 4, 9 and 10
Symbol Description
T Test item temperature
T
AMBIENT
Ambient temperature
T
NO-max
Maximum non-operating temperature (highest design temperature for the equipment to
survive not powered)
T
NO-min
Minimum non-operating temperature (lowest design temperature for the equipment to
survive not powered)
T
SU-high
Maximum start-up temperature (highest design temperature of the equipment, at which the
equipment can be switched on)
T
SU-low
Minimum start-up temperature (lowest design temperature of the equipment, at which the
equipment can be switched on)
T
Q-max
Maximum qualification temperature (highest design temperature at which the equipment
demonstrates full design ability)
T
Q-min
Minimum qualification temperature (the lowest design temperature at which the equipment
demonstrates full design ability)
P Pressure
MODE 1 Functionally inert (test item not energized). Normally applicable to the non-operating
condition.
MODE 2 Partially functioning. Conditions as detailed in applicable design specifications, but normally
applicable to conditions during launch.
MODE 3 Fully functioning (test item fully energized and fully stimulated). Normally applicable to
conditions during orbit.
Initial and final “functional and performance test”
Intermediate reduced functional and performance test
t
E
Dwell time
Switch-on (Start-up)
Switch-off
Table 7.1 – Legend and symbols for thermal vacuum (and cycling) test sequences (Fig.
7.3) [17]
ECSS
15 February 2002
ECSS--E--10--03A
72
5.1.16.4 Thermal cycling test temperatures
The qualification temperature limits shall be in accordance with subclause
5.1.15.5.
5.1.16.5 Thermal cycling test cycles and duration
The equipment shall be tested in the thermal cycling test sequence performing
8 thermal cycles, as shown in Figure 9.
NOTE Characteristic parameters of the thermal cycling test are
given in Table 16.
a. The first thermal cycle begins after the equipment is functionally tested at
ambient temperature.
b. The equipment shall be switched off and the temperature increased up to the
high non-operating level (T
NO--max
).
c. After a dwell time t
E
, the temperature shall be decreased to the maximum
(hot) start-up level (T
SU--high
) and then the temperature shall be stabilized at
the high operating temperature (T
Q--max
).
d. After the time t
E
, the functional and performance test shall be performed.
e. The equipment shall be switched off and the temperature shall be decreased
and maintained at the minimum non-operating temperature (T
NO--min
) dur-
ing a time t
E
.
f. The temperature shall be increased to the minimum(cold) start-up tempera-
ture and the equipment switched on.
g. When stabilized at the lowoperating level (T
Q--min
), and after the time t
E
, the
functional test shall be performed.
h. The equipment shall be cycled between T
Q--max
and T
Q--min
until the number
of cycles specified in Table 16 is achieved.
i. During the last cycle, the equipment shall be functionally tested at T
Q--max
and T
Q--min
.
j. At the end of last cycle, the temperature shall be raised to ambient conditions
and the final functional and performance test performed.
k. The temperature rate of change < 20 ºC/min shall apply only to equipment
within the space vehicle. For equipment outside the space vehicle, higher
gradients are specified in the appropriate equipment specifications.
Table 16: Thermal cycling test parameters (qualification)
Parameter Condition / Value
Start cycle Hot
n (number of cycles) 8
t
E
(dwell time at T
hot
/T
cold
)
a, b
2 h
dT/dt (temperature rate of change) < 20 ºC/min
Stabilization criterion 1 ºC/h
a
T
hot
= T
Q-max
or T
NO-max
.
b
T
cold
= T
Q-min
or T
NO-min
.
Table 7.2 – Thermal cycling test parameters (qualification) (Fig. 7.3) [17]
Levels Duration
Qualification
margins
Acceptance
margins
Qualification Acceptance
10

C 5

C 8 cycles 4 cycles
Table 7.3 – Qualification and acceptance test levels and durations according to
ECSS-E-10-03A standard [17]
All the previous curves and tables comes from standards considering
conventional spacecraft. Regarding OUFTI-1, some modifications will be
provided to meet the particular CubeSat thermal design.
Lionel Jacques
2
nd
Master in Aerospace Engineering
106 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
OUFTI-1’s TMM results and operating/non operating (NO) tempera-
ture ranges showed that the batteries were the more critical components.
Their NO temperature range, identical to the discharge one is only -20

C
to +60

C while their operating one is reduced to 0

C to +45

C. On another
hand, all the other units (structural and electronic) have a wider operating
(and non-operating because they are identical) range at least equal to the
-40

C to +85

C industrial temperature range. The dissipation transistor
whose temperature range is even wider: from -65 to +200

C (operating or
non-operating).
Predicted (op) Qualification Non-operating Start-up
Min Max Min Max Min Max Min Max
Batteries 5

C 35

C 0

C 45

C -20

C 60

C 5

C 35

C
Other -30

C 35

C -40

C 45

C -45

C 85

C - -
Table 7.4 – OUFTI-1’s test temperatures definition
Table 7.4 confirms that the batteries are well the more restrictive com-
ponents. The predicted temperature of the structure ranges from -30

C
to +35

C but measures have been taken to ensure that the temperature
of the batteries does go beyond their limits. Thanks to the active and pas-
sive thermal control system, the batteries should now range between +5

C
and +35

C. The non-operating lower temperature of the batteries (-20

C)
is thus higher than the operating temperature of the CubeSat (-30

C): the
temperature of the switched off CubeSat could only decrease to -20

C
while if it is switched on, the heater are designed to maintain the batter-
ies above 0

C when the temperature of the structure is -30

C. This is in
contradiction with the above figure 7.3
Two solution are proposed to deal with this problem:
1. The first solution is therefore to keep the CubeSat switched on af-
ter the first functional test and decreasing the temperature down
to the low qualification operating temperature T
Q,min
instead of the
non-operating one. The CubeSat should also be operational (at least
the heaters) during all the subsequent cycles. The number of cycles
should thus take into account the level of charge of the batteries.
A separate test should be performed on the batteries for their non-
operating qualification.
2. The second solution consist in decreasing the temperature down to
the non-operating one of the batteries during the first cycle with the
CubeSat switched off. Then raise the temperature to the low start-
up temperature T
SU,low
before switch on the CubeSat and decrease
its temperature again to the operating one. Here again, the CubeSat
should be kept switched on to allow the heating system to maintain
the batteries above 5

C and the number of cycles should be defined
by taking into account the level of discharge of the batteries resulting
from the heaters consumption.
Lionel Jacques
2
nd
Master in Aerospace Engineering
107 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
7.3 Test set-up
Thermal testing involves specific equipments such as vacuum cham-
bers allowing active thermal control. Moreover, vibrations testing which
are not discussed in the present thesis (but in the STRU subsystem [42])
are also required and usually performed before thermal testing. Both will
be performed at the Centre Spatial de Liège (CSL) which is one of the
four Coordinated European Environmental Test Facilities (CETeF) of the
European Space Agency (ESA). CSL has six vacuum chambers called FO-
CAL (Facility of Optical Calibration at Liège) whose diameter range from
0.25 to 6.5m (with 1.5m, 2m, 3m and 5m). While the satellite Planck have
been tested in FOCAL 5 during nearly two months, the foreseen vacuum
chamber for OUFTI-1 thermal testing is FOCAL 1.5.
Nevertheless, many test parameters have still to be defined in addition
to the different temperature ranges of the thermal vacuum cycling. The
following questions need to be answered:
How is fastened the CubeSat in the vacuum chamber?
How to impose the temperature variation on the CubeSat?
How the low pressure levels are achieved and is-it harmful to the
CubeSat?
Is solar simulation feasible?
What are the functional test that will be performed and how to col-
lect the data?
Shall the batteries be charged during the tests?
. . .
While some of these questions will remain partially unanswered be-
cause the overall design is not yet defined, some answers can already be
given.
As explained in test specification section, the required pressure level
is 10
5
hPa. Actually, the pressure required depends on the size of the
vacuum chamber: the bigger the chamber, the lower the required pres-
sure. The key parameter is to ensure that the mean free path of remaining
molecules in the chamber is greater than the chamber characteristic size
(diameter). The pumping phase is divided in two stages: a first one per-
formed through classical pump down to 10
2
mbar and the second using
a turbomolecular pump to decrease the pressure down to 10
5
mbar. De-
velopment tests demonstrate that such pressure levels were harmful to the
batteries that inflates under mechanical stresses.
The first two questions are not independent because the way a satellite
or instrument has its temperature imposed influence the way it is fas-
tened to the chamber and conversely. Each thermal tests set-up has its
own dedicated set-up and heating/cooling method. There is no universal
method. Nevertheless different heating and cooling methods are available
each with its own advantages and drawbacks. They can be classified in
two categories: radiative and conductive methods and are presented in
Table 7.5
Lionel Jacques
2
nd
Master in Aerospace Engineering
108 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
Method Advantages Drawbacks
R
a
d
i
a
t
i
v
e
Solar simu-
lation
- Does not assume a
prior known environ-
ment
- Solar simulation
chamber not available
at CSL
- Accurately simulates
solar environment
- Cannot simulate non-
solar heat loads
- Allow GMM error de-
tection
- Parallel illumination
required large set-up
Heating
lamps
- lamps can be placed
efficiently and are in-
dependent, providing
good flexibility
- many lamps required
and interfering with
each other
Heater
plates
- environment accu-
rately known and al-
low independent sur-
face control
- requires knowledge
of absorbed fluxes and
cooling in heater plates
C
o
n
d
u
c
t
i
v
e
heater - good for appendages
such as antennas,
booms etc.
- requires knowledge
of absorbed fluxes and
test blanket are re-
quired of heaters are
mounted to them
heater
plates
- direct and indepen-
dent surface control
- limited flexibility
Table 7.5 – Thermal testing heating/cooling methods [24]
Solar simulation would have been appropriate to simulate the real ther-
mal loads on the CubeSat and would have allowed the use of the solar
cells. However solar simulation is difficult to implement because the heat
generated by the halogen is difficult to cope with unless the lamps are rel-
atively far from the other cooler environment elements. It is not available
for the moment at CSL.
However, because of its cubic shape, it seems convenient to use of
heater plates, one for each face of the CubeSat. Moreover, the continuous
rotation of the CubeSat involves relative uniform external temperatures of
the sides and we can assume that all the faces have the same temperature.
All the heater plates surrounding the CubeSat can then be controlled by a
unique fluid loop. The fluid loop consist of a nitrogen thermally controlled
inlet line that impose the temperature of the plates. The heater plates as-
sembly is covered with MLI to reduce exchanges and losses with the walls
of the chamber remaining at ambient temperature. The plates and MLI
constitute the so-called thermal shroud. Figure 7.4 shows the shroud used
for Planck testing in FOCAL 5. The internal sides of the shroud are panted
in black (here it is black opened honeycomb) to enhance exchanges be-
tween the satellite and the shroud. Sometimes, thermal straps between the
satellite and the shroud are used to enhance heat exchange and increase
the heating or cooling phase. However, seeing the relative low thermal
inertia of the CubeSat, such measure should not be necessary.
The CubeSat has five of its six faces covered with solar cells. Conse-
Lionel Jacques
2
nd
Master in Aerospace Engineering
109 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
Figure 7.4 – Example of thermal shroud used for the testing of Planck in FOCAL 5
quently, the only remaining face available to fasten the CubeSat in the
chamber is the antennas’ face. Again, the mounting interface should be
insulating from the chamber walls ambient temperature. Figure 7.5 illus-
trates the foreseen set-up with fluid loop controlling the shroud.

Thermal vacuum
chamber

Temperature
controlled shroud

CubeSat

Mounting
interface

Temperature
controlled fluid
loop

Reference point

Data acquisition
path and control
TC/TM

Figure 7.5 – Thermal vacuum cycling test set-up
Figure 7.5 also shows that a way to communicate with the CubeSat and
collect data must be provided. At any time, all the parameters must be
available to allow a comprehensive understanding, monitoring and con-
Lionel Jacques
2
nd
Master in Aerospace Engineering
110 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
trol of the CubeSat behavior. The thermal cycling curves presented above
require to define a reference point temperature that is representative of the
mean temperature level of the equipment. A temperature sensor shall be
located at the reference point. During the test, the shroud temperature are
controlled by the operator in such a way that the reference point temper-
ature follows the thermal cycling specification curve. In the present case,
two reference points are defined and located as follows:
1. on the structure. This reference point shall be representative of the
global temperature of the CubeSat. The sensor must obviously not
be located on solar cells nor anodized rails. Therefore, the more con-
venient place is again the antennas’ face (face 4). However, it shall be
placed far enough from the mounting interface.
2. the second reference point is the batteries. As the batteries will al-
ready be equipped with temperature sensors, these sensors will be
used as reference point to monitor the batteries temperature. In any
case the temperature of the batteries would go beyond their limits,
proper measures shall be taken to bring it back within its safe range.
Concerning the performance and functional tests that are performed
during the first (TB) and last cycles, they are difficult to be clearly defined
at this development phase. However, we can already assume that the fol-
lowing tests should probably be conducted:
Each defined mode should be tested. Among these, full D-STAR
transmitting and receiving. As the CubeSat is inside the vacuum
chamber, communication through the antennas are useless and suit-
able testing procedures shall therefore be defined without the use
of the antennas. Moreover, the thermal vacuum cycling test will be
performed in stowed configuration. Nevertheless, during transmis-
sion, the amplifier shall be used normally and its temperature shall
be carefully monitored especially during the hot case.
Extensive beacon functional and performance tests shall be con-
ducted in hot and cold cases.
The dissipation system shall be switched on with full power. If nec-
essary, additional power is provided through the USB interface to
meet the predicted dissipated power with 10% margins [? ]. The
dissipation transistor and batteries temperatures shall carefully be
monitored.
In the cold case, the heating system of the batteries shall demonstrate
that the design implementation fully meets the requirements.
All the housekeeping parameters shall be monitored and saved, es-
pecially the temperatures to allow the correlation of the Thermal
Model.
As a comprehensive test is not possible, separate tests should be con-
ducted, under ambient pressure:
Lionel Jacques
2
nd
Master in Aerospace Engineering
111 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Chapter 7. Testing
Simulations have shown that the antennas could undergo relatively
high temperatures (up to more than 200

C at their free extremity).
For this reasons, specific functional tests should be conducted in the
deployed configuration with dedicated thin heaters located on the
antennas’ free extremity. Heaters should be designed to cope with
the convective losses.
7.4 Summary
Thermal testing is a compulsory step in the development of a satellite.
Concerning OUFTI-1, the tests shall be conducted at the Centre Spatial de
Liège. As the CubeSat development is not terminated, a detailed test de-
scription was tot possible. However, foreseen set-up and functional tests
have been described. Once the CubeSat will be fully designed, a test pro-
cedure document shall be written in agreement with the Centre Spatial de
Liège and all the subsystems to carefully define the functional tests that
should be performed on the CubeSat.
Lionel Jacques
2
nd
Master in Aerospace Engineering
112 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Conclusions
Throughout this work, different thermal models with different pur-
poses and complexity have been carried out. The first models brought to
light that issues would occur in both cold and hot cases. Nevertheless,
only the Detailed Thermal Model enabled a comprehensive investigation
of these issues. As the thermo-optical properties of the structure and ther-
mal properties of the batteries were subjected to some relatively high un-
certainties, measurements have also been conducted through which the
heat capacity and transverse conductivity of a typical LiPo battery were
determined, as well as the emissivities of the two coating of the aluminum
frame. Moreover, contact phenomenon in the structure have been con-
firmed through IR imaging. Those results were therefore included in the
DTM.
It has been presented that both hot and cold cases had a strong impact
on the design of the CubeSat.
In the cold case, simulations showed that the batteries would undergo
too low temperatures and that they were out of their temperature ranges
when coming out of the eclipse. A heating system has therefore been de-
signed: two 250 mW Minco heaters, set in parallel combined with three
temperature sensors (to avoid the problem encountered by Compass-1),
will ensure that the temperature of the batteries will remain within its
allowable range. The heaters will be switched on when the temperature
reached 5

C. To reduce losses and increase its efficiency, the DTM showed
that insulating the batteries is required. Covering the batteries with a low
emissive tape such as aluminum thermal tape is planned and Nylon wash-
ers combined to titanium spacers shall be used to fasten batteries’ PCB.
Even in the worst case, we verified that the consumption of the heating
system remains acceptable compared to the capacity of the batteries: up
to 150 mAh compared to the 100 mAh (and therefore 2000 mAh for both
batteries) typical capacity of one LiPo battery. However, as the battery
model is not yet fully defined, the heaters have not yet been ordered, de-
pending on the dimensions of the batteries. On another hand, we contact
the company Moss Plastic Parts to order some sample nylon washers be-
cause creep phenomenon should be carefully studied to ensure structural
integrity of the batteries’ PCB fastening.
Concerning the hot case, the dissipation system developed by EPS sub-
system was the cause of nearly all issues: involving too high temperatures
on the batteries and the dissipation transistor. To solve these problems,
drastic measures had to be taken in order to reduce temperatures: we
showed that a thermal strap encircling the dissipation transistor would be
required and fastened to antennas’ deployment mechanism panel, involv-
ing a 10

C temperature decrease of the batteries (from 48

C to 38

C) and
20

C of the transistor (from 115

C to 95

C) . In addition to that, relocating
113
Conclusions
a part of the dissipated power through two Minco resistances (reference
XHK5377R26.3L12B) in parallel also fastened to the antennas’ panel in-
volves another 3

C and 30

C temperature decrease of the batteries and
transistor, respectively reduced to 35

C and 67

C. Finally, for reliability
purposes, batteries’ PCB should be fastened to EPS2 PCB instead of EPS
one because it contains less dissipative components and is therefore cooler.
The Minco resistances have been ordered to evaluate the feasibility of their
integration on antennas’ panel, which has been confirmed.
The third issues occurring in the hot case involves the COM amplifier,
because of its quite low efficiency and therefore high heat dissipation.
No concrete measures have been taken to solve this problem because too
many uncertainties still surround its definition. However, we stated that
the amplifier should be localized either close to one corner of the COM
PCB or it the center of the PCB. locating the amplifier on one edge should
be avoided. Furthermore, its efficiency should be taken into account when
selecting the amplifier. For instance, using the ADL5541 could reduce the
temperature from 70

C to 45

C because the ADL5541 only dissipates
0.5 W instead of 1.75 W assumed for classical amplifiers.
On the educational point of view, working on this project has been
very enriching and its primary goal has been achieved: through it, we
gained an invaluable hands-on experience in satellite design, especially in
thermal design. Working within a CubeSat project has numerous advan-
tages. Among them, in our opinion, working within a team is the main
one and has been extremely constructive, it allows a global view of the
project and highlights the strong interaction that may occur between the
different subsystems.
Moreover, this project allowed us to present the project in different
international events and workshops during which we met other CubeSat
teams and exchanged ideas. Namely, we participated to the Second Euro-
pean CubeSat Workshop at ESTEC, we presented the project on the Space
Days event in Liège and at the Redu Eurospace Center.
CubeSat project also allows to work within an industrial framework
and to collaborate with famous companies or agencies such as the Eu-
ropean Space Agency (ESA), EADS Astrium, Thales Alenia Space, Azur
Space and many others.
Concerning the future developments, numerous tests should be con-
ducted to correlate the model and because this is the only way to ensure
that the design meets the specification requirements. Particularly, the heat-
ing system should be extensively tested. Tests should also be conducted
on the heat rejection design to verify that the strap and relocated resis-
tances play their role. The effect of hot antennas should be investigated
to see the effect on their radiation pattern because the vacuum tests that
will be performed et the Centre Spatial de Liège will probably be in the
stowed configuration.
At last, according to Prof. Bob Twiggs, the CubeSat leitmotiv should
be:
Test, test and test again !
Lionel Jacques
2
nd
Master in Aerospace Engineering
114 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Appendix
115
Appendix
A Acronyms
ADCS Attitude Determination and Control System
A&M Department of Aerospace and Mechanical Engineering
BOL Beginning Of Life
COTS Commercial-Off-The-Shelf
CSL Centre Spatial de Liège
D-STAR Digital Smart Technologies for Amateur Radio
DTM Detailed Thermal Model
EOL End Of Life
EECS Department of Electrical Engineering and Computer Science
EPS Electrical Power Supply
ESA European Space Agency
ESATAN European Space Agency Thermal Analysis Network
ESTEC European Space Technology Research Center
FE Finite Element
FEM Finite Element Model
FHTS Fluid Heat Transport System
FOCAL Facility for Optical Calibration at Liege
GMM Geometric Mathematical Model
GND Ground
IC Integrated Circuit
MCRT Monte Carlo ray-tracing
MLI Multi-layer Insulation
OBC On-Board Computer
OUFTI Orbiting Utility For Telecommunication Innovation
P-POD Poly-Picosatellite Orbital Deployer
PCB Printed Circuit Board
REF Radiative Exchange Factor
SSETI Student Space Exploration and Technology Initiative
STM Simplified Thermal Model
STRU Structure & Configuration
TCS Thermal Control System
TMM Thermal Mathematical Model
ULg University of Liège
OBC PCB main on-board computer printed circuit board
OBC2 PCB secondary on-board computer printed circuit board
EPS PCB main electrical power supply printed circuit board
BAT PCB secondary printed circuit board on which the batteries are attached
EPS2 PCB innovative electrical power supply printed circuit board
COM PCB communication subsytem on-board computer printed circuit board
Lionel Jacques
2
nd
Master in Aerospace Engineering
116 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Appendix
B Aluminium frame faces links
This appendix explains how the links between the different nodes of
the frame have been computed. The method is the same as the one used
for the computation of the global links in the high level model : by impos-
ing the temperature gradient across the desired geometry, a SamcefField
finite element analyses gives the resulting flux from which the thermal
conductance can be deduced. The thermal resistance/conductance does
only depends on the geometry and conductivity of the frame links.
Face 3
Figure 6 describes the resistances network (in red) representing the heat
flow path between the nodes of the face. For symmetry reasons, there are
only two different resistances, R
1
along the diagonal and R
2
along the
horizontal/vertical.


2

1

1


1


1

Figure 6 – Conductance model of face 3
R
1
and R
2
are easily determined: the first one is obtained by consid-
ering a quarter of the face and setting the boundary conditions on both
extremities. By this way, the diagonal does not get involved in the con-
duction path (figure 7, left) and R
1
is equal to one half of the difference of
temperatures divided by the resulting flux:
R
1
=
1
GL
=
∆T
2Q
tot
=
0.5
0.04094
= 12.21 [K/W]
Then, R
2
is obtained by involving the diagonal in the flow path and
setting one of the two temperature at its extremity (figure 7, right). The
equivalent resistance is therefore equal to:
Lionel Jacques
2
nd
Master in Aerospace Engineering
117 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Appendix
R
eq
= R
1
+ R
1
//R
2
= R
1
+
R
1
R
2
R
1
+ R
2
R
2
=
R
1
(R
1
R
eq
)
R
eq
2R
1
One finally gets: R
2
= 33.29 [K/W] with R
eq
= 21.14 [K/W]
Figure 7 – Face 3 conduction SamcefField results
The previous model assumed perfect contact on all the bracket surface
but experiments of chapter 6 showed the presence of contact phenomenon.
Hence, the heat flow passing from face 3 to face 1,2,5 or 4 shrinks around
the contact area, close to the screw. To take into account constriction effect
around the screw hole, an additional resistance R
c
(in blue) is considered
between the nearby node and the screw hole. To determine this constric-
tion resistance, the boundary conditions were set around the screw holes.
The computed flux correspond to the green resistances R

1
of figure 6.
R
c
can then be approximated by differentiating R
1
and R

1
. This leads to
R
C
= R

1
R
1
= 15.02 12.21 = 2.81 [K/W].
Finally, after some simplifications, the total transverse resistance, from
one side to the opposite one is equal to:
R
tot
= R
c
+ R
1
+ R
1
//R
2
= 2.81 +21.14 = 23.95 [K/W]
while the finite elements global resistance, between to opposite screw
holes, is 23.76 [K/W]
Faces 1, 2 and 5
Face 1, 2 and 5 are identical and but no longer symmetrical. This means
that instead of two different resistances, there are three ones: R
1
, R
2
and
R
3
as shown on figure 8.
Using the superposition principle, one can perform three FE simula-
tions to obtain a set of three equations for the three unknowns: one for
R
1
+ R
2
, the second for R
1
+ R
3
and the third one for R
2
+ R
3
. One finally
gets R
1
= 24.55 [K/W], R
2
= 24.62 [K/W] and R
3
= 37.04 [K/W].
Face 6
Face 6 conductances are identical to face 3 except for the link with the
additional node 3610, representing the bracket fastened to face 4.
Lionel Jacques
2
nd
Master in Aerospace Engineering
118 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Appendix

3

1

2

Figure 8 – Conductance model of faces 1, 2 and 5
Face 4
Face 4 links have all been determined by the same way, through FE mod-
els.
All the conductance are summarized in Table 6.
nodes GL nodes GL nodes GL nodes GL
Face 1, 2 & 5 - 3100, 3200 & 3500 series
1-2 4.07 3-2 4.07 5-4 4.06 7-6 4.07
1-9 2.69 3-9 2.69 5-9 2.69 7-9 2.69
1-8 4.06 3-4 4.06 5-6 4.07 7-8 4.06
Face 3 - 3300 series
1-2 8.19 3-2 8.19 5-4 8.19 7-6 8.19
1-9 3.00 3-9 3.00 5-9 3.00 7-9 3.00
1-8 8.19 3-4 8.19 5-6 8.19 7-8 8.19
Face 4 - 3400 series
1-2 5.50 3-2 10.8 5-4 4.06 7-6 8.19
1-9 3.17 3-9 5.56 5-9 2.69 7-9 3.00
1-8 64.9 3-4 3.85 5-6 4.07 7-8 8.19
Face 6 - 3600 series
1-2 8.19 3-2 8.19 5-4 8.19 7-6 8.19
1-9 3.00 3-9 3.00 5-9 3.00 7-9 3.00
1-8 8.19 3-4 8.19 5-6 8.19 7-8 8.19
2-10 62.7
Table 6 – Aluminium frame faces conductances, in [W/K] 10
−2
Lionel Jacques
2
nd
Master in Aerospace Engineering
119 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Appendix
C Analytical computation of VF between parallel
rectangles
Here is the formula used for the view factor computation in the ver-
ification process of DTM’s Geometrical Mathematical Model. The view
factors between each elements of the two opposite PCBs were indeed re-
quired in the Gebhart’s formula. Figure ?? illustrates the different notation
used.

2

1

1

2

2

1

1

2

1

1



Figure 9 – View factor between two parallel rectangles [15]
F =
1
(x
2
x
1
)(y
2
y
1
)
2

i=1
2

j=1
2

k=1
2

l=1
( 1)
i+j+k+l
G(x
i
, y
j
, η
k
, ξ
l
, z)
where the function
G(x, y, η, ξ, z) =
a + b c

with
a = (y η)
_
(x ξ)
2
+ z
2
arctan
_
y η
_
(x ξ)
2
+ z
2
_
b = (x ξ)
_
(y η)
2
+ z
2
arctan
_
x ξ
_
(y η)
2
+ z
2
_
c =
z
2
2
log
_
(x ξ)
2
+ (y η)
2
+ z
2
_
The view factors with the lateral aluminum panels representing the
lateral frame were indirectly deduced from the reciprocity properties of
the view factors (A
i
F
ij
= A
j
F
ji
) and from the energy conservation (∑F
ij
=
1 for a closed system) that is enforced by setting the aluminum panels such
as they form a closed environment with the PCBs.
Lionel Jacques
2
nd
Master in Aerospace Engineering
120 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Appendix
D Advanced Simulink Thermal Model
Here are the advanced Simulink model diagrams. First, the global one:
on the left is the heat fluxes computation module already presented and
on the left the thermal lumped parameter model which is described in the
second diagram. Within this, the PCB stack block is detailed in the third
and last module. The results obtained with this model are relatively well
correlated with ESATAN/ESARAD ones.
T_Faces
9
P_TOT
8
P_ELEC
7
T_COM
6
T_EPS2
5
T_EPS
4
T_Bat
3
T_OBC
2
T_OBC2
1
°K to °C6
u+b
°K to °C5
u+b
°K to °C4
u+b
°K to °C3
u+b
°K to °C2
u+b
°K to °C1
u+b
°K to °C
u+b
Thermal Model
HF1
HF2
HF3
HF4
HF5
HF6
P_COM
P_EPS2
P_EPS
P_OBC2
P_OBC
P_Bat
T_OBC2
T_OBC
T_Bat
T_EPS
T_EPS2
T_COM
Incident Heat fluxes
DCM
r_sat _ECI
r_sun_ECI
Cs
albedo_coeff
T_Earth
HF1
HF2
HF3
HF4
HF5
HF6
P_ELEC
Electrical _Power _Face1
Total _Power _Face6
Total _Power _Face5
Total _Power _Face4
Total _Power _Face3
Total _Power _Face2
DCM
t
Total _Power _Face1
P_ELEC
T_Face6
T_Face5
T_Face4
T_Face3
T_Face2
Electrical _Power _Face6
Electrical _Power _Face3
Electrical _Power _Face5
Electrical _Power _Face2
r_eci
T_Face1
JD
r_sun
Cs
sun
0
0
0
0
0
0
0
0
0
0
255
0.3
Figure 10 – Advanced Simulink thermal model main high level diagram
Lionel Jacques
2
nd
Master in Aerospace Engineering
121 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Appendix
T_COM
6
T_EPS2
5
T_EPS
4
T_Bat
3
T_OBC
2
T_OBC2
1
Solver
Configuration
f(x)=0
PS S
PS S
PS S
PS S
PS S
PCBs Stack
P_COM
P_EPS 2
P_EPS
P_OBC 2
P_OBC
P_Bat
T_BAT
T_COM
T_EPS 2
T_EPS
T_OBC 2
T_OBC
A
B
T
A
B
T
A
B
T
A
B
T
A
B
T
Deep_Space_Temp
GL 6/5
A B
GL 6/4
A
B
GL 6/2
A
B
GL 6/1
A B
GL 4/5
A
B
GL 3/6
A
B
GL 3/5
A
B
GL 3/4
A B
GL 3/3
A
B
GL 3/2
A B
GL 3/1
A
B
GL 2/4
A B
GL 2/1
A
B
GL 1/5
A B
Face 6
H
e
a
t F
lu
x
e
s
T
e
m
p
F
a
c
e

6
Face 5
Heat Fluxes Temp Face 5
Face 4
Heat Fluxes Temp Face 4
Face 3
H
e
a
t F
lu
x
e
s
T
e
m
p
F
a
c
e

3
Face 2
Heat Fluxes Temp Face 2
Face 1
Heat Fluxes Temp Face 1
Deep Space
Temperature
3
P_Bat
12
P_OBC
11
P_OBC2
10
P_EPS
9
P_EPS2
8
P_COM
7
HF6
6
HF5
5
HF4
4
HF3
3
HF2
2
HF1
1
Figure 11 – Advanced Simulink thermal model low level diagram: global model
Lionel Jacques
2
nd
Master in Aerospace Engineering
122 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Appendix
T_BAT
1
T_OBC
5
T_OBC2
4
T_EPS 3
T_EPS2 2
T_COM
1
PS S
PS S
PS S
PS S
PS S
PS S
Switch
A
B
A B
A B
A B
A
B
A
B
A
B
S
A
B
S
A
B
S
A
B
S
A
B
S
A
B
S
A
B
PS S
PS S
OBC2
OBC
A
B T
A
B T
0.5
EPS2
EPS
0
A
B
A
B
A
B
A B
A B
A B
A
B
A
B
A
B
A
B
A
B
A
B
A
B
COM
Battery 2 top
Battery 2 bottom
Battery 1 top
Battery 1 bottom
BAT
P_Bat
6
P_OBC
5
P_OBC2
4
P_EPS
3
P_EPS2
2
P_COM
1
Figure 12 – Advanced Simulink thermal model low level diagram: PCB stack model
Lionel Jacques
2
nd
Master in Aerospace Engineering
123 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Appendix
E 2D conduction lumped parameter method
The lumped parameter method allows to transform differential equa-
tions into algebraic equation. Concerning thermal conduction, the alge-
braic equation are moreover linear (if materials properties do not depend
on temperature) and can therefore be written under a matrix form. Here,
the case a 2D plate will be considered but the method can easily be gen-
eralized to 3D problems.
The lumped parameter method relies on a discretization of the do-
main into smaller isothermal elements called nodes. For rectangular ele-
ments, Figure 13 illustrates the incoming fluxes exchanged between adja-
cent nodes and energy conservation leads to:
Q
1,n
+ Q
2,n
+ Q
3,n
+ Q
4,n
+ Q
n
= ρc

T
n
In case of steady state,

T
n
= 0 and using electrical analogy, one has:
GL
x
(T
1
T
n
) + GL
x
(T
3
T
n
) + GL
y
(T
2
T
n
) + GL
y
(T
4
T
n
) = Q
n
where GL
x
=
kt∆
y

x
and GL
y
kt∆
x

y
are respectively the thermal conduc-
tances along x and y directions with k the thermal conductivity and t the
thickness of the plate.

3 1
4
2

Δ

Δ

Figure 13 – Equivalent conductivity of BAT PCB fastening in function of the PA
washer thickness and thermal conductivity of the spacer
Generalizing for n nodes the equations can be set under the following
matrix form
A T = g
where A is a n n matrix, A(i, j) representing the conductive coupling
between nodes i and j and g(i) the external loads of node i.
If there are B temperature boundary conditions and therefore I = n
B unknowns, the system can be divided as follows:
T =
_
¸
¸
¸
¸
¸
¸
¸
¸
_
T
1
.
.
.
T
I
T
I+1
.
.
.
T
n
_
¸
¸
¸
¸
¸
¸
¸
¸
_
=
_
T
I
T
B
_
Lionel Jacques
2
nd
Master in Aerospace Engineering
124 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Appendix
Splitting matrix A and vector g in the same way, this gives:
_
A
I I
A
IB
A
BI
A
BB
_ _
T
I
T
B
_
=
_
g
I
g
B
_
and the unknowns temperature T
I
are deduced without inversing the
entire A matrix:
T
I
= A
−1
I I
_
g
I
− A
IB
T
B
_
The method has been correlated with SamcefField has shown in Figure
14.
Figure 14 – Equivalent conductivity of BAT PCB fastening in function of the PA
washer thickness and thermal conductivity of the spacer
Lionel Jacques
2
nd
Master in Aerospace Engineering
125 University of Liège
Applied Sciences Faculty
Academic Year 2008 −2009
Bibliography
[1] I. Alet, Contrôle thermique des engins spatiaux. École nationale
Supérieure de l’Aéronautique de l’Espace, 1998. (Quoted page 23.)
[2] ESARAD
TM
User Manual - UM-ESARAD-024, Alstom Aerospace, Cam-
bridge Road, Whetstone, Leicester LE8 6LH United Kingdom, oct
2007. (Quoted pages ix et 43.)
[3] ESATAN
TM
User Manual - UM-ESATAN-004, Alstom Aerospace, Cam-
bridge Road, Whetstone, Leicester LE8 6LH United Kingdom, oct
2007.
[4] Analog Devices, Inc., “50 mhz to 6 ghz rf/if gain block, adl 5541,”
2007, revision 0. (Quoted pages 86 et 99.)
[5] ——, “High performance narrow-band transceiver ic data sheet, adf
7021,” 2007, revision A. (Quoted page 21.)
[6] Vega User’s Manual, Issue 3, revision 0, http://www.arianespace.
com, Arianespace, Boulevard de l’Europe, B.P. 177, 91006 Evry-
Courcouronnes cedex, France, mar 2006.
[7] Azur Space Solar Power, “Azur space solar power gmbh,” http://
www.azurspace.com. (Quoted pages xii, 5, 42, 47 et 75.)
[8] H. W. Babel, C. Jones, and K. David, “Design properties for state-of-
the-art thermal control materials for manned space vehicles in Leo,”
Acta Astronautica, vol. 39, pp. 369–379, Sept. 1996. (Quoted page 42.)
[9] V. Beukelaers, “From mission analysis to space flight simulation of
the OUFTI-1 nanosatellite,” Master’s thesis, University of Liège, Bel-
gium, 2009. (Quoted pages viii, 9, 23 et 76.)
[10] L. Chiarello, “Design and implementation of the control and moni-
toring system for the ground station of nanosatellite OUFTI-1 of Uni-
versity of Liège,” Master’s thesis, University of Liège, Belgium, 2009.
(Quoted page 8.)
[11] J. H. Chou, “Design of Thermal Control Sub-System (TCS), Picosat
course,” May 2007. (Quoted page 42.)
[12] “Compass-1 Phase B documentation,” http://www.raumfahrt.
fh-aachen.de/, Jun 2004. (Quoted page 18.)
[13] “Delfi-C3’s website,” http://www.delfic3.nl/. (Quoted page 19.)
126
Bibliography
[14] DuPont
TM
, “Summary of Properties for Kapton
R
Polyimide
Films,” www2.dupont.com/Kapton/en_US/assets/downloads/
pdf/summaryofprop.pdf. (Quoted pages xii et 47.)
[15] J. R. Ehlert and T. Smith, “View Factors for Perpendicular and Par-
allel, Rectangular Plates,” Journal of Thermophysics and Heat Transfer
1993, vol. 7, pp. 173–174, 1993. (Quoted pages xi et 120.)
[16] ESA-ESTEC, “Spacecraft thermal control design data, PSS-03-108,”
1989.
[17] “ECSS-E-10-03A - space engineering : Testing,” ESA Publications Di-
vision ESTEC, Feb 2002. (Quoted pages x, xii, 104, 105 et 106.)
[18] “ECSS-Q-70-71A - space product assurance : Data for selection of
space materials and processes, rev.1,” ESA Publications Division ES-
TEC, Jun 2004. (Quoted pages xii, 47 et 48.)
[19] N. Evrard, “Développement de l’infrastructure de mesure du
nanosatellite OUFTI-1,” Master’s thesis, Gramme Institute, Belgium,
2009. (Quoted pages viii et 10.)
[20] H. Flemming, “DTU Satellite Systems and Design Course, Cubesat
Thermal Design,” http://www.dsri.dk/roemer/pub/cubesat, Sep
2001. (Quoted page 42.)
[21] ThermaCAM
TM
Researcher User’s manual, FLIR Systems, Dec 2006.
(Quoted pages ix, 42, 61 et 63.)
[22] P. Fortescue, J. Stark, and G. Swinerd, Spacecraft Systems Engineering,
3rd ed. Wiley, March 2003.
[23] S. Galli, “Mission design for the cubesat oufti-1,” Master’s thesis, Uni-
versity of Liège, 2008.
[24] D. Gilmore, Spacecraft Thermal Control Handbook: Fundamental Technolo-
gies, 2nd ed. Aerospace Press, December 2002. (Quoted pages viii,
ix, x, xii, 16, 23, 25, 42, 70, 71, 104 et 109.)
[25] S. Hannay, “Modeling of the Attitude Determination and Control
System of the nanosatellite OUFTI-1’,” Master’s thesis, University of
Liège, Belgium, 2009. (Quoted pages viii, 7, 32 et 52.)
[26] J. Hardy, “Implémentation du protocole AX.25 à bord du nanosatel-
lite OUFTI-1,” Master’s thesis, Gramme Institute, Belgium, 2009.
(Quoted page 10.)
[27] P. G. Harkness, “An aerostable drag-sail device for the deorbit and
disposal of sub-tonne, low Earth orbit spacecraft,” Ph.D. dissertation,
Cranfield University, School of Engineering, 2006. (Quoted page 42.)
[28] R. Henrard, “Mission Design for the CubeSat OUFTI-1,” Master’s the-
sis, University of Liège, Belgium, 2009. (Quoted pages viii, 8 et 76.)
Lionel Jacques
2
nd
Master in Aerospace Engineering
127 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Bibliography
[29] L. Jacques, “Rapport de stage : Etude thermique d’un capteur
cylindro-parabolique,” University of Liège - Centre Spatial de Liège,
Belgium, Tech. Rep., nov 2008.
[30] L. Jacques and G. Pierlot, “Spécifications mécaniques des cartes pc104
destinées au projet oufti-1,” Department of Aerospace and Mechani-
cal Engineering, University of Liège, Liège, Belgium, Tech. Rep., mar
2009.
[31] Kokam Co., Ltd., “Superior Lithium Polymer Battery, specification
for SLPB554374H,” Mar 2007. (Quoted page 21.)
[32] ——, “Superior Lithium Polymer Battery, specification for
SLPB723870H4,” Apr 2007. (Quoted pages x et 87.)
[33] P. Ledent, “Design and Implementation of On-board Digitaly Con-
troled Electrical Power Supply of Student Nanosatellite OUFTI-1 of
University of Liege,” Master’s thesis, University of Liège, Belgium,
2009. (Quoted pages 8, 50, 77 et 87.)
[34] ——, “Test report - Vacuum tests for Varta and Kokam batteries,”
University of Liège, Belgium, Tech. Rep., May 2009. (Quoted pages x,
92 et 93.)
[35] Y. Madhour, “Phase C - SwissCube thermal Analysis,” http://
swisscube.epfl.ch/, EPFL - LTCM, Lausanne, Switzerland, Tech. Rep.,
Jun 2008. (Quoted page 49.)
[36] F. Mahy, “Design and Implementation of On-board Telecommuni-
cation System of Student Nanosatellite OUFTI-1 of University of
Liège,” Master’s thesis, University of Liège, Belgium, 2009. (Quoted
pages viii, 8 et 76.)
[37] Microchip Technology, Inc., “Pic18f2331/2431/4331/4431 data
sheet,” 2003. (Quoted page 21.)
[38] “Military specification - MIL-A-8625F Anodic Coatings, for Alu-
minum and Aluminum Alloys, with Amendment 1,” http://www.
wbdg.org/ccb/FEDMIL/a8625.pdf, Sep 2003. (Quoted page 60.)
[39] “Military specification - MIL-DTL-5541F Chemical Conversion coat-
ings on Aluminum and Aluminum alloys,” http://www.wbdg.org/
ccb/FEDMIL/dtl5541f.pdf, Nov 1990. (Quoted page 60.)
[40] Minco, “Flexible Heaters Design Guide,” www.minco.com. (Quoted
pages 93 et 94.)
[41] PC/104 Embedded Consortium, “PC/104-Plus Specification, version
2.3,” www.pc104.org, University of Liège, Belgium, Tech. Rep., Oct
2008. (Quoted page 48.)
[42] G. Pierlot, “OUFTI-1 : Flight system configuration and struc-
tural analysis,” Master’s thesis, University of Liège, Belgium, 2009.
(Quoted pages viii, 10 et 108.)
Lionel Jacques
2
nd
Master in Aerospace Engineering
128 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Bibliography
[43] P. Poinas, “Rosetta - Experiment Thermal Interface Mathematical
Models,” ESA - ESTEC - TOS-MCT, Tech. Rep., Jun 1999, issue 2,
reference : RO-EST-TN-3021. (Quoted page 42.)
[44] ——, “Satellite Thermal Control Engineering, prepared for "SME"
2004,” Jun 2004. (Quoted page 16.)
[45] P. Prat, “GLAST Large Area Telescope Calorimeter Subsystem Ther-
mal Design and Analysis,” Laboratoire Leprince-Ringuet, Ecole poly-
technique, Tech. Rep., Mar 2003. (Quoted page 42.)
[46] F. Primore, “Online calculator for the thermal properties in the plate
direction for multi-layer glass-epoxy PCBs,” http://www.frigprim.
com/online/cond_pcb.html. (Quoted page 47.)
[47] Pumpkin, Inc., “CubeSat Kit User Manual,” http://www.cubesatkit.
com, 2005. (Quoted pages 6 et 21.)
[48] ——, “CubeSat Kit
TM
FM430 Flight Module Datasheet,” Jun 2008, re-
vision J. (Quoted page 21.)
[49] A. Rajendra, “Pressure Sensor Development Using Hard Anodized
Aluminum Diaphragm and Thin Film Strain Gauges,” Master’s the-
sis, Indian Institute of Science, Department of Instrumentation, Ban-
galore, Apr 2006. (Quoted page 42.)
[50] P. Rochus, “Conception d’expériences spatiales,” 2008. (Quoted
pages 16, 42 et 70.)
[51] Semelab, “2N5151 High Speed Medium Voltage Switches Datasheet,”
mar 1997. (Quoted page 69.)
[52] Sheldahl Red Book - Aerospace Thermal Control Materials, http://www.
sheldahl.com, Sheldahl, 1150 Sheldahl Road, Northfield, MN 55057,
oct 2007. (Quoted pages 63 et 88.)
[53] “SwissCube’s website,” http://swisscube.epfl.ch/. (Quoted page 18.)
[54] D. Teney, “Design and implementation of on-board processor and
software of student nanosatellite OUFTI-1 of University of Liège,”
Master’s thesis, University of Liège, Belgium, 2009. (Quoted page 10.)
[55] P. Thirion, “Mission Design for the CubeSat OUFTI-1,” Master’s the-
sis, University of Liège, Belgium, 2009. (Quoted pages 8, 50 et 87.)
[56] D. Vallado, Fundamentals of Astrodynamics and Applications , 3rd ed.
Microcosm Press and Kluwer Academic Publishers, 2007. (Quoted
page 23.)
[57] J. Van Noord, “Icarus Final Thermal Report,” http://data.engin.
umich.edu/Icarus/, Nov 1999. (Quoted page 42.)
[58] VARTA Microbattery GmbH, “VARTA Sales Program and Techni-
cal Handbook, VARTA PoLiFlex
R
Batteries Superior Polymer Tech-
nology,” http://www.varta-microbattery.com, Feb 2008. (Quoted
page 21.)
Lionel Jacques
2
nd
Master in Aerospace Engineering
129 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009
Bibliography
[59] P. Veughs and G. Schmetz, “ESARAD - ESATAN, Introduction aux
logiciels, Rapport SF-258,” University of Liège, Belgium, Tech. Rep.,
May 2007. (Quoted page 47.)
[60] P. Villano, M. Carewska, and S. Passerini, “Specific heat
capacity of lithium polymer battery components,” Ther-
mochimica Acta, vol. 402, no. 1-2, pp. 219 – 224, 2003.
[Online]. Available: http://www.sciencedirect.com/science/article/
B6THV-47NF62S-4/2/4651f95f8ef619f68976a9d85dfe9b60 (Quoted
page 59.)
[61] N. T. Wakelyn, “Thermal control coating,” National Aeronautics And
Space Administration, NASA, Tech. Rep., nov 1964, patented Jul,
1968. (Quoted pages ix et 60.)
[62] J. R. Wertz and W. J. Larson, Space Mission Analysis and Desig, 3rd ed.
Microcosm Press and Kluwer Academic Publishers, 1999. (Quoted
page 21.)
[63] J. Wertz, “Conception et réalisation du système de déploiement des
antennes du nanosatellite OUFTI-1,” Master’s thesis, Gramme Insti-
tute, Belgium, 2009. (Quoted pages viii, 9 et 70.)
[64] A. Wiberg, “Thermal study on hardware with comparisons between
physical measurements and simulations,” ABB Automation Tech-
nologies AB, Tech. Rep., may 2005. (Quoted page 42.)
Lionel Jacques
2
nd
Master in Aerospace Engineering
130 University of Liège
Applied Sciences Faculty
Academic Year 2008 2009

1

Space isn’t remote at all. It’s only an hour’s drive away, if your car could go straight upwards.

Sir Fred Hoyle

ii

Acknowledgments

I

wish to thank all the people who helped me, during the whole year, to carry out this master thesis.

I express my sincere thanks to Prof. Pierre Rochus for his precious advice and encouragement and for giving me the opportunity to enter the very interesting world of spacecraft thermal control. I am greatly indebted to Tanguy Thibert for his guidance, suggestions, useful feedbacks and all the support he has given throughout my work. I expressed also my sincere gratitude to all the people working at the Centre Spatial de Liège for their availability and support in helping me conducting the different tests. I also wish to thank Philippe Poinas for his valuable advice and help in developing OUFTI-1’s thermal model. My special thanks goes to all the OUFTI-1 student and management team, especially to Amandine Denis and Jonathan Pisane who managed the project with an incomparable team-spirit. Particularly, I am thankful to Prof. Gaetan Kerschen for his valuable suggestions and recommendations and to Vincent Beukelaers who helped me during the tests. Last but not least I thank my family and I thank you, Géraldine, for encouraging, supporting me and not being weary of always hearing me talking about OUFTI-1. . . Again, many thanks to all of you, for keeping our eyes turned toward the sky. . .

Harzé, June 15, 2009.

iii

This present thesis focuses on the thermal design of OUFTI-1 whose goal is to guarantee all components are functioning within their allowable temperature range. The two other payloads are high efficiency solar cells provided by Azur Space and an innovative electrical power system developed with Thales Alenia Space ETCA. This student project takes place within the framework of a long-term goal program called LEODIUM (Liège in Latin).Abstract OUFTI-1 is the first nanosatellite developed at the University of Liège and even the first one ever made in Belgium. based on their results. OUFTI-1 will hopefully be launched on the new European launcher Vega with eight other student nanosatellites. CubeSat. ESARAD] iv . OUFTI-1 will be the first satellite ever equipped with one of the latest digital amateur radio communication protocol : the D-STAR protocol. Keywords : [ OUFTI-1. With this in mind. ESATAN. proper measures will be taken to ensure all the components works indeed in their allowable range of temperatures. The goal of this program is to provide hands-on experience to engineering students through the development of a series of nanosatellite for scientific experiments in cooperation with space industries of the region of Liège. LEODIUM. satellite. different thermal model of increasing complexity are developed within both Matlab/Simulink and ESATAN/ESARAD environments. This technology represents one of the three payload of OUFTI-1. Thermal. since one target of the mission is to test this new protocol in space.

. . . . . .2 3. .2 Cold Case . . 3. . . . . . . . 2. .1 The Project Story . . . . . .3 2. The Subsystems . . . . . . . . . . . . . . . . . . . . . . . . Advanced Simulink Model . . . . . Thermal environment related to OUFTI-1’s orbit 3. . . . . . . . . . . .1 Conduction . . . . . . . . . . . . . . .2 Solar flux . . . . .1 1. . . .2 Radiation . . . . . . . . . . . . . . . 3. . . . . . . . .4 3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3. . . . . . . . . . . . . .6 Results . . . .1 Implementation . . . . . . . . Thermal control systems of other CubeSats 2. . . . . . . . .5. . . . . . 2. . . . . . . . . . . . . . . . . . .5. . . . . . . . . . . . . . . . . . . . . . . . .3 3 Preliminary thermal analysis Notations . . .4 3. .4. . . . . . mission and nanosatellite 1. . . . . . . . . . . . . . . . . . .5 3. . . .4 1. . . . . . . . . . . . . . . . . Cyclic Transient Model . . . . . . . . . Sensitivity analysis . . . . .2 1. . . . . . . . . . . . . . . . . . .1. . .3. . . . .3 Albedo flux . . 3. . . . . . . . . . . . . . 3. . . . . . . . . . . . . .3.4 Earth infrared flux . . . . . . .2 Validation . . . . . . . . . . . . . . . . . . . . . .3. . . v viii xii 1 2 2 4 5 7 11 13 13 13 13 16 17 18 20 20 21 22 22 23 23 26 27 27 29 29 30 31 33 33 34 35 2 The Thermal Control Subsystem Space heat transfer . The Satellite . . . . . . . .Contents Contents List of Figures List of Tables Thesis outline 1 OUFTI-1 project. . . . The Mission . . . . . . . . .5 2. . . . . . . . . . . . . . . .3. . . . . . 3. . . . . The thermal environment . . . . . . . . . . . . . . . . . . Thermal Requirements . . . . OUFTI-1’s anatomy . . . . . . . . . . . . . . . . . . . . . . . . why Thermal Control ? and how ? . . 3. . . . . . . . . . . . . . . . . . . . . . . .1 Hot Case .3. . . . . . .5 Aerothermal flux . . . . . . .6 3. . . . 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 The orbit . . . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 3. .7 . . . . . 3. . . . . . . .4.3 1. . 3. . . . . . . . . . . . . . . . . . . 3. . . . . . . . Thus. . . .2 2.

. .4. . . . . .3.3 4. .3. . . . 4. . . . . . . . . . . .1 The structure . .2 Thermo-optical properties . . . . . . . . . . . . . .3 Hot Case 1 . . . . . . . . . . . .2 TMM . . . . Results . . . . . . . . . . . . . . . . . .4 6. . .2. . . . . . . . . . . . . . . Worst cases definition . . . . . Results . . . aluminum frame emissivity . . . . . .5. . . . . . . . 38 38 40 40 40 41 42 44 44 46 47 48 49 50 50 53 54 56 56 56 56 57 58 59 59 60 62 63 64 67 67 67 70 70 70 70 74 77 77 78 80 80 82 82 84 85 86 94 5 Measurements 5.1 Nodal breakdown . . . 6. .4. . . . . . . . . . . . . . . . .2 Hot case issues . . . . . . . .4 4. . . . . . . . . . Thermal Mathematical Model . . . . . . . . . . . . . . 5.1 GMM . . . Geometric Mathematical Model 6. . 5. 6. . . . . .2. . 4. . . . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermal Mathematical Model . . . . . . . . . . . . . . . 4. . . . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5. . . .3 Measuring emissivity with a thermographic camera Experimental setup . . . . . . . . . . . . . . . . . . 5. . . . . . . .3. . 6. . . . . .2 The model . . . . . . . .1 aluminum frame contact resistance . . 5. . . 5. . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . .4 Simplified Thermal Model 4. . . . . . . . . . . .3 The PCBs stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 4. . . . .1 Hot case .2 Internal power . 4. . . . . .1. . . . . . . . . . . .2. . . . . . . . . . . . . . . .1 Nodal breakdown . . . . . . . . . . . . . . .2. . . . . . . . . . . .5. . .2. . . . . . . . . 6 Detailed Thermal Model 6. . . . .4 The batteries . . . . . . . . . .4 6. . . .5. . . . . . . . . . . . . . .4. . . . . . . . .5 . . . 6. . . .3 Orbit & attitude . . . . . . . . . . . . . .5. . . . . . . . . .1 Cold case issues . . . . .3.1 5. . . . . 6. . .3 ESARAD considerations . . Checking the model . .2 Cold Case . . 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. . . . . 6. . 4. . . . . . . . . . . 4.1 4. . . . . . . . . . . . .2 Thermo-optical properties . . . . . . 6.1 Anodizing vs Alodine . . . . 4. . . . .3. . . . 6. . . . . . . . . . . . 6. . . . . . . . . . .3. . . . . . . . . . . . . . . . . . . . . 5.3 Model adjustment & results . . . . . . . .1 Absorbed powers . . . . . . . . . . . . . . . . . . .2 The solar cells . . .5 . . . . .1 Conductive network . . . . . . . . . 6. . . 4. .3 Sensitivity analysis .1. . . . . . . . . . . . . . . . . 5. . . . . . . . . . . . .3. Why and which measurements ? The Battery . . . . . . . . . .2. . .3. . . . . . . . . . . . . . . . .2 Cold case . . . . . . . . . . . . . . . . . . . .3. . . . . . . . . . . . 4. . . . . . . . . . . . . . . . 4. . . . . .3. . . . . . .4 Hot Case 2 . . . . . . . . .4.2 5. . .2 Thermal modeling with ESATAN & ESARAD Geometric Mathematical Model . . . . . . . . . . . . .2 . . . . . . . . . .4 Radiative coupling .2 6. .3 5. . . . . . . . . . . Parametric analysis and design . . . . . . .1 Experimental setup . . 6. . . . . .3 6. . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .7 Testing 7. . .3 7. . . . . . Summary . . . . . Advanced Simulink Thermal Model . . . . . . . . . . 120 . . . . . . . . . . . . . . .4 Test and model philosophy Tests specifications . . . . . . . . . . . . . . . . . . . . . . . . . 2D conduction lumped parameter method . . . . . . . . . 116 . . . . . . . . . . . . . Analytical computation of VF between parallel angles . . . 124 126 Bibliography . . . . 121 . . . . . . . Test set-up . . . . . . . . . . . . . 115 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 7. . . . . . . . . . . . . . . . . . . 101 101 102 108 112 113 Conclusions Appendix A B C D E Acronyms . . Aluminium frame faces links . . . . . . . . . .1 7. . . . . . 117 rect. .

. . . . . .9 3. . .4 1. . . . . .8 3. . Fitting of the Gilmore abacus curves with the formula 3. .2 3. . . . . . . . . . . . . . . . . . . . Incident albedo irradiation on a surface element in Earth orbit [24] . . . . . 8 OUFTI-1’s antenna deployment mechanism panel[63] .2 . . . . .2 Thermal model processing flow chart . . . 8 Main EPS engineering model . Nominal Earth IR and Albedo fluxes seen by a surface pointing to the Earth on OUFTI-1’s orbit . . . . . . .13 3. .7 3. .5 1. . .List of Figures 1. . . . . . . . . . . . . . . . . . . . . the CubeSat Kit Structure. . . . . Classical orbital elements . Cold Case temperature evolution .14 4. .7 1. . . Hot Case definition. . .3 1. . . . . Simulink two-body propagator for incident flux computation : main window (top) and Albedo and IR calculation box (bottom) . . . . . . . 6 On the left. . . . 9 1. 12 2. . . . . . . . . . . . . .4 3. . . . . . . . . .8 Mission description : Doppler-Effect compensated zones . . . . . . Tmax & Tmin sensitivity to aluminum panel thermo-optical properties . .1 3. . . . . . .10 Finite element modal analysis with SamcefField[42] . . . . . . . . . 14 17 22 24 25 26 27 28 28 30 31 33 34 35 36 36 3. . . on the right. . . . . . . . .1 2. . Infrared fluxes incident on each face and effective area ratio Effective area ratio . . . . . . . . . .2 1. . . . . .12 3. . . . . . . . influence of orbit orientation . . . . . . . . . . . . . . . . . . 7 OUFTI-1’s test RX PCB[36]. . . .1 4. . .10 3. . 39 Geometric Mathematical Model nodal breakdown in ESARAD 41 viii . . . Cyclic transient Matlab Simulink solver flow chart . . . . . . . 5 OUFTI-1’s three payloads . . . . Advanced simulink thermal model cold case results . . . . . . . . Advanced simulink thermal model hot case results . . . .5 3. . . . . . . . .2 Planck’s black body radiation curves for different temperatures . . . Thermal environment of a satellite in LEO . . . . . . . . . . . . . . . . . . . Beukelaers in Simulink[9] . . . . . . . . .3 3. . . . 10 1. .11 3. . . . . 10 1. . . . . . . .6 1. . . . .6 3. . . . . . . . 9 Snapshot of the Simulator developed by V. . . . . . . .11 OUFTI-1 exploded view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 1. . . . . . . . . . . the P-POD 7 Earth’s magnetic field orientation on the orbit[25] . . . . . . . . . . . . . . . . . . . . . . . .9 OUFTI-1’s second OBC PCB[19] . . . Validation of the cyclic transient Matlab/Simulink solver with ESATAN/ESARAD . . . . .[28] . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . face 6 and 1. . . 5. . . . . . .8 4. . . . . . . the CubeSat with one aluminum panel before integration. . . . . . TRW and Lockheed Martin Bolted-Joint Resistance Data [24] Thermal conductance between two thin plates of different section . . . . . . Influence of the EPS-BAT spacers conductivity on batteries’ temperature . . . . . . . . . . . . . On the left. . . . . . . . . . . . . . . . . . . . . . . . . 5. . . . . . . . . . . . . . . . . . . . . . . 5. .4. . . . . . . . . . . . . . . . . . . .2 5. Hot case absorbed power . .4 Evolution of the emissivity and absorptivity of aluminum sample subjected to a conversion coating surface treatment in function of the reaction time [61] . Evaluation of the conductive link of aluminum frame through a finite element analysis within SamcefField . . . . . . . . . . . . . .7 4. . . . . . . . . . Conductive link computation through a finite element analysis within SamcefField of a midplane standoff . . . . . . . . . 1: Surroundings. . . . 5. . . .11 4. . . . .3 6. . 6. . Cold case temperatures . . Cold case absorbed power . . . . . . . 5. . . . . . . . . . . . Radiative exchange factor convergence with number of rays fired (MCRT) .6 Influence of the object temperature and emissivity for a thermographic measurement . .9 aluminum frame emissivity test results . . . . . . . . . . Battery test comparison between measurements and adjusted model . . . . . . . 3: Atmosphere. . . .14 4. .2 6.3 4. . . . . . . . . . . . fastened with one screw . . 5. . . . . . . . . .5 4. .12 4. . . .9 4. . . . . . . . . . Hot case temperatures . . Detailed Thermal Model nodal breakdown of the aluminum frame . . . . . 5. Evolution of the temperatures during one orbit for a rotation rate of 5 deg/s . . . 5. . . . . . . . . . . . . . . . . .3 On the left. . . . . . . . . . . .15 5. . . . . . . . . . . . . . . . On the right. Geometric Mathematical Model in ESARAD . . . . . . . .7 aluminum frame emissivity test setup . .10 4. . . . . . 2: Object. . . . . . . . . . . . . . . . . . . . . . . 60 61 62 63 63 64 64 65 68 68 71 71 72 73 . . . . . . . . Solar cell power balance . . . . . . . from ESARAD User Manual . . . . . .6 Midplane standoffs fastening to the aluminum frame . 4: Camera . . MCRT convergence from ESARAD user manual [2] . . . . . . . .11 Infrared images Battery test setup . . . . . . . . . . .13 4. . .5 A schematic representation of the general thermographic measurement situation from the ThermaCAM User Manual [21]. . . face 6 and 4.4 6. . .5 6. . . . TM Battery test Simscape model . . . . . . . . . . . . . . . .4 4. . . . . . . . . . .1 6. . . . . . . . . . . . . . . . . . . . 41 43 43 44 45 46 49 51 51 52 53 54 55 57 58 59 Battery test setup . . . . . .10 Aluminum frame contacts in front of bright background. . .1 5. .6 4. . . . fastened with 3 screws and on the right. Planet centered Local Orbit Coordinate System (LOCS). . external layout and corresponding thermal conductance . . . . . . . . . . . . .8 Infrared capture of the structure heated up to 90 C . the panel integrated with the cells by EADS Astrium . . . . . .

. . . . . . .26 6. . . Thermal margin terminology for NASA/JPL programs [24] Thermal margin terminology from ECSS-E-10-03A standard [17] . . . . . . . . . . . . . . . . . . . . .15 6. . . . . . . . . . . . . . . . . . . . . . . . . . Thermal vacuum cycling test sequence [17] . . . . . . .1 7. . .24 6. . . Effect of temperature and aging on solar cells efficiency . . . . . . . . . .16 6. . . . Heating system control flow chart . . . . Cold case absorbed power . . batteries’ capacity decrease (worst voltage U=2. check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9 6. . Hot case 2 heat flow map . . . . . . Hot case 1: Hot spot due to the dissipation transistor and temperature distribution on the antennas . . . . . . . . . . . . . . . . . . . . . . . Amplifier temperature increase in function of its relative location on the COM PCB . . . . . . . . . . . . . . . . .21 6.18 6. . . . . . . . . . . Check of the conductive links through the bus and equivalent resistance model . . . . . . . Thermal strap (angle bracket in red) design on engineering EPS PCB (left) and its fastening configuration with the relocated resistances on antenna panel (CATIA Model on the right) . . . . . the inter-PCBs conductive model . . . . . . . . . . . .17 6. . . . . . . Temperature of a SLB 603870H KOKAM battery during the cold test [34] . . . ESARAD radiative coupling check for one PCB node (left: analytical. . . . . . . . Hot case 2: Hot spots due to the dissipation transistor and amplifier on COM PCB . . . . .23 6. . . . . . . . . . . . . . . . Cold case absorbed power .30 7. . . . . . . . . . . . . . . . . . . . . . .27 6. . . . . . . . . . . . . . . . . . . . . . . . . . .14 6. . . . Checking the conductive links inside PCBs through isotherm shape . . .13 6. . . . . . . . . . . . . Example of thermal shroud used for the testing of Planck in FOCAL 5 . . . . . . . . Effect of insulating the batteries on the heater required power for a given threshold [5 C] . Equivalent conductivity of BAT PCB fastening in function of the PA washer thickness and thermal conductivity of the spacer . . . . . . Hot case absorbed power . . . On the right. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 On the left. right: MCRT) . . . . . . .25 6. . .10 6. . . . .29 6. .28 6. . . . . . .7 6. . . . . . . . .11 6. . . . .19 6. . . . . .2 7. . . . . . . . Evolution of the battery minimal temperature. . .3 7. . . . . . . . . . . . . . Battery temperature profile of a typical LiPo battery. . . . . . . . . . . . . . .22 6. . . . . . . . . . . . . . . . . . . . . . . . . . Power distribution in the equivalent resistance and the transistor of the dissipation system . . . . . . . .12 6. . . . . . . . . . . . . .6. . . . . . .20 6. . . . . . . . . . . . . . . .8 6. . . . . . . . . . . . . . . . . . .5 V) and time the heater is turned on in function of the heater power for different thresholds . here the KOKAM SLPB723870H4 [32] . . inserted copper washers in EPS2 engineering model. . . . Batteries’ PCB fastening concept . New design hot case heat flow map . . . . . . 74 76 78 79 79 80 81 81 82 83 84 85 86 87 88 89 90 90 92 93 94 95 97 98 104 104 105 110 . . . . . Hot case 1 heat flow map .

. . . . . . . . . . View factor between two parallel rectangles [15] . . . . . . . . . . . . . . . . . . . . Advanced Simulink thermal model low level diagram: PCB stack model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conductance model of faces 1. . . . . . . . . . . . . . . . . . . Advanced Simulink thermal model main high level diagram Advanced Simulink thermal model low level diagram: global model . . . . . . . . . Equivalent conductivity of BAT PCB fastening in function of the PA washer thickness and thermal conductivity of the spacer . . . . . . .5 6 7 8 9 10 11 12 13 14 Thermal vacuum cycling test set-up . Face 3 conduction SamcefField results . . 110 117 118 119 120 121 122 123 124 125 . . . . . . . . . . . . . . . . . . .7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conductance model of face 3 . 2 and 5 . Equivalent conductivity of BAT PCB fastening in function of the PA washer thickness and thermal conductivity of the spacer . . . . . . . . . . . . . . . . . . .

.1 4. . . . . . . . . . . . . . . . .2 3. . . . . . OUFTI-1’s test temperatures definition . . 7.4 7. . . .6 4. . . for CS = 1367 [W/m2 ] . . . . . . .1 3.2 7. . . . . . .2 6. . . . . . . Thermal testing heating/cooling methods [24] . . . . . . . . Effects of the different solutions to the hot case issues . . . . . . . . . . . . . . . . . . . OUFTI-1’s Thermal Requirements . . . . . . . .3) [17] Qualification and acceptance test levels and durations according to ECSS-E-10-03A standard [17] . . . . . TM Thermal conductivity of Kapton [14] and RTV S691 adhesive [18] . . . . . . .5 4. . .4 3. . . . . . . . . OUFTI-1s thermal heat capacity . . . . Equivalent conductance between aluminum frame faces . . xii . . . . . . . . . . . . . . . . . DTM Hot Cases internal dissipation . . . . . . . . . Comparison of hot case temperatures before and after modifications . . . . . . . . .1 6. . . . . . . 7. . . . . . . . .4 4. . . . . . . . Aluminium frame faces conductances. . . . . . . . . . Sensitivity coefficients to thermo optical properties . .2 4. .1 7. . . . . . . . . . . . . . . . . . . . . . . . . . . PCBs links materials thermal conductivities . . . . . . . . . . . 17 21 31 32 34 35 42 45 45 46 47 48 49 75 77 96 98 106 106 106 107 109 119 Thermo Optical Properties . . . in [W/K ] 10 2 . . . . Preliminary Hot & Cold case definition . . . . . . . .List of Tables 2. . . . . . . Nodal capacitances . . PCBs links . . . .1 3. aluminum alloys properties . . . . . . . . . . . . . . . .4 7. . . . . .3 7. . . . . . . . . . . . . . .7 6. . . . . . . .3) [17] . . . . . . . . . . . . . Thermal cycling test parameters (qualification) (Fig. . . . . .3 4. . . . . . . . 28% solar cells efficiency parameters [7]. . . . . External area ratios . . . . . . . . . . . . . . . . . . . .5 6 Thermal control means . . . . Legend and symbols for thermal vacuum (and cycling) test sequences (Fig. .3 6. . . . .3 3. . . . . . . .5 4. . . . . . . . . . . .

two ESATAN/ESARAD are developed: a first one to confirm the results of the single node Matlab/Simulink model and a second one.Thesis outline This thesis focuses on the thermal analysis of the nanosatellite OUFTI1. Then a preliminary analysis is conducted through Matlab/Simulink softwares to have a first guess of the temperatures the CubeSat will undergo. Through it. a series of parametric analysis are performed to enable suitable design modification accordingly with the other subsystems to enforce thermal reliability with sufficient margins. thermal vacuum and cycling tests that will be performed next year at the Centre Spatial de Liège will introduced. far more detailed. Finally. developed at the University of Liège. to reveal and/or confirm the possible issues. With this detailed model. the mission and the CubeSat concept. Then we will draw conclusions and propose future development. This work will be divided into 7 parts : first. Before developing this detailed model. the thermal environment the satellite will be exposed to is studied. measurements will be performed on critical elements such as the battery or the main structure. The ultimate goal of this thesis is to take suitable measures to ensure all the components will be functioning in their safe range of temperatures and a proper heat rejection. 1 . a brief introduction about the project. After that.

proposed to test a new amateur radio digital technology in space on board of a CubeSat : the D-STAR protocol. besides meaning Liège in Latin. Pierre Rochus.1 The Project Story The LEODIUM program The OUFTI-1 project takes place within the framework of a long-term goal program called LEODIUM. the University of Liège. through the LEODIUM program. In fact. mission and nanosatellite 1 1. stands also for Lancement En Orbite de Démonstations Innovantes d’une Université Multidisciplinaire (Launch into Orbit of Innovative Demonstrations of a Multidisciplinary University). Mr. In 2005.OUFTI-1 project. Yet it needed a name : OUFTI-1. Last year. standing for Orbiting Utility For Telecommunication Innovation. which is a typical expression of the city of Liège. One must wait until September 2007 when Mr. Luc Halbach. was charged with the training of students to the design of miniaturized satellites. has now the ambitious goal to develop other CubeSats to keep on giving students satellite hands-on experience. why not ? The idea of developing a nanosatellite was already in the mind of many people at ULg but yet nothing concrete had began. a consortium of space industries and research centers in the Liège region. 2 . Different possibilities were foreseen and the project finally began with the participation in the Student Space Exploration and Technology Initiative (SSETI) of the European Space Agency. as the president of Liège Espace and Deputy General Manager for Space Instrumentation of the Centre Spatial de Liège. For two years. It did not even take one month for a team of students and professors to set up around the newborn project. students were involved in the European Student Earth Orbiter (ESEO) with the design of the solar panels deployment system and also in the development of the the Narrow Angle Camera (NAC) for the European Student Moon Orbiter (ESMO). It involves both the University of Liège and Liège Espace. Stefania Galli and Jonathan Pisane performed respectively the feasibility study and a detailed analysis of the D-STAR protocol. in their master theses. sales manager of Spacebel. A CubeSat. LEODIUM.

the ESA Education Office in cooperation with the Directorate of Legal Affairs and External Relations and the Vega Programme Office in the Directorate of Launchers. Jacques Verly (EECS Dept. Goliat for Bucharest in Roumania and SwissCube for Lausanne in Switzerland) and there are also two back-up (HiNCube for Narvik in Norway and UWE-3 for Würzburg in Germany). Prof. PW-Sat for Warsaw in Poland. AtmoCube for Trieste and e-st@r for Turin in Italia.)) and Luc Halbach from Spacebel. Computer engineering. Initially. Amandine Denis presented the project at the 5th Annual CubeSat Developers’ Workshop in San Luis Obispo.).). 7 ULg. the status of our project has been presented during the 2nd European CubeSat Workshop in January 2009 at ESTEC and at the 6th Annual CubeSat Developers’ Workshop in San Luis Obispo. We finally got a positive answer in June 2008. ESA published a call for proposal for CubeSat on board of Vega to which we submitted our own proposal in March 2008. This year. Eight other CubeSats are selected (Xatcobeo for Vigo/INTA in Spain. Electrical engineering. three professors (Prof. UNICubeSat for Rome. In April 2008. 4 ULg. After presenting the project at the Vega Maiden Flight CubeSat Workshop at the European Space Research and Technology Center (ESTEC) in January 2008. Gaëtan Kerschen (A&M Dept. Prof. Electrical engineering. the Vega Maiden Flight was scheduled for November 2009. Pierre Rochus (CSL and A&M Dept.Chapter 1. Electrical engineering. Computer science. thanks to the good presentation made 6 months earlier. 2 Isil Institute. 5 Gramme Institute. : : : : : : : : : Samuel Hannay1 Renaud Henrard2 and Francois Mahy3 Philippe Ledent3 and Pierre Thirion3 Laurent Chiarello4 Jérôme Wertz5 Vincent Beukelaers1 Nicolas Evrard6 . USA. Johan Hardy6 and Damien Teney7 Gauthier Pierlot1 Lionel Jacques1 Aerospace engineering. mission and nanosatellite Vega Maiden Flight In October 2007. 1 first master student (Philippe Ledent) two graduate students (Amandine Denis and Jean-François Vandenrijt). more students joined the project so that we are now 13 students divided up into 9 subsystems: ADCS COM EPS GND MECH MIAS OBC STRU THER 1 ULg. 6 Gramme Institute. OUFTI-1 project. USA. Robusta for Montpellier in France. This tight schedule involved some decisions that will be described later. issued a first Announcement Opportunity offering a free launch on the Vega maiden flight. The Team The initial team was composed of 2 second master students (Stefania Galli and Jonathan Pisane). the flight has recently been delayed to spring 2010. Mechanical engineering. Nevertheless. This year again. 3 ULg. This left us only a little bit more than one year to develop our CubeSat. Lionel Jacques 2nd Master in Aerospace Engineering 3 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

Chapter 1. OUFTI-1 project, mission and nanosatellite

Management is carried out by Amandine Denis and Jonathan Pisane, both PhD students at the University of Liège.

1.2

The Mission
In addition to give students satellite hands-on experience, OUFTI-1 mission has three main objectives. OUFTI-1 will be the first satellite ever equipped with D-STAR. The primary goal is thus the space qualification of this new technology so that a functional D-STAR repeater can be given to the amateur radio community. Hence, we will use the amateur radio frequency bandwidth reserved for the communications between the satellite and the ground station. Furthermore, seeing the worldwide coverage of the amateur radio community, they will reveal to be an invaluable help us to keep our nanosatellite in good health when it is not in sight of our ground station. But what is D-STAR? As previously said, it is a new amateur-radio communication protocol. The main difference between conventional protocols and D-STAR is that it is digital. The quality on the voice is then definitely better. Another D-STAR key feature is that it allows simultaneous data and voice transmission. What becomes D-STAR in space? In fact, using D-STAR in space have some consequences. The frequency shift due to Doppler Effect during one pass of the satellite is indeed too large for the acceptable bandwidth of the ICOM 2820 transceiver Doppler Effect compensation capabilities. Unfortunately, this transceiver is, at the moment, the only one able to deal with D-STAR protocol available on the market. Doppler Effect will so have to be on-board compensated. Seeing that, there will be two system-selected doppler-compensated coverage zones : the first around ULg for control and another one dynamically determined. Concerning the second one, there will probably be a registration procedure on our website for the user to reserved one pass of the satellite over a specific region. This is illustrated on the figure 1.1. This zone definition fits well with an emergency usage of D-STAR in space : in case of large scale disasters (Katrina Hurricane or more recently the earthquake that hits center Italy), common communication networks are often unavailable. D-STAR is then very useful because this recently developed protocol allows not only digital voice communication but also data transmission such as GPS data at the same time! The disaster zone could then be selected to be Doppler compensated and allow amateur radio to use this D-STAR satellite repeater to communicate and send GPS data. Anyway, as it is also shown on the figure 1.1, any user who has the ability to make his own Doppler compensation could technically communicate with the CubeSat when it is in sight from the user and in D-STAR mode (and not in AX.25 mode for Telecommand/Telemetry communication). As previously said, OUFTI-1 has two other payloads : high efficiency solar cells and an innovative EPS. As it will be presented in the next paragraph, nanosatellite are ideal low cost solutions for testing new technologies. Concerning OUFTI-1, AZUR SPACE Solar Power GmbH proposed us to test their new solar cells. These are 30% efficiency triple junction
Lionel Jacques 2nd Master in Aerospace Engineering

4

University of Liège Applied Sciences Faculty Academic Year 2008 2009

Chapter 1. OUFTI-1 project, mission and nanosatellite

• 2 system-selected dopplercompensated coverage zones - ULg for control - Dynamically determined

• Personnal doppler – compensation possible within OUFTI-1 coverage zone

Figure 1.1 – Mission description : Doppler-Effect compensated zones

GaAs cells, compared to their previous 28% model or their silicon cells (type S-32) having only an efficiency of 17%. The cells are thus going to be tested in real conditions. AZUR SPACE is a world recognized company in space solar cells production and nowadays, more than 300 satellites have been equipped with solar cells without any failure[7]. The third payload is a new digital electrical power unit developed in cooperation with Thales Alenia Space (ETCA) about which we will come back later. This innovative EPS is digitally controlled and based on a PIC microcontroller and other components such as planar transformers. When the batteries voltage is high enough (and the CubeSat normally works), digital EPS will supply and be connected to the 3.3V power bus. Yet, this EPS has a relatively lower efficiency than the analog 3.3V converter (50% compared to 90%).

1.3

The Satellite
Up to this point, even if the word CubeSat has been used at least ten times, it has not yet been presented. But what really is a CubeSat or a nanosatellite ? In fact, for several years satellite have become larger and larger (e.g. Envisat launched in 2002 which weights more than eight tons and is 26 meters tall or Hubble in 1990 with its 11 tons and 16 meters) but such missions required generally about ten years of development and billions of euros. This is why there is a new tendency in reducing the size and thus costs and development time. Actually, there is now a classification of miniaturized satellites according to their wet mass : minisatellites (between 100 and 500 kg), microsatellites (between 10 and 100 kg), nanosatellites (between 1 and 10 kg) and finaly Picosatellites (between 0.1 and 1 kg). During the Second European Workshop at ESTEC, Prof.
Lionel Jacques 2nd Master in Aerospace Engineering

5

University of Liège Applied Sciences Faculty Academic Year 2008 − 2009

Chapter 1. OUFTI-1 project, mission and nanosatellite

Figure 1.2 – OUFTI-1’s three payloads

Bob Twiggs from the Department of Aeronautics & Astronautics at Stanford University presented the concept of even smaller satellites : femtosatellites, weighting less than 100g, for teaching purposes ! Miniaturized and nanosatellites are now very sought-after by industrials and academics for testing new technologies under real conditions at low cost and short development time. Within this framework, Prof. Bob Twiggs originally proposed the 10x10x10cm 1kg CubeSat Standard. Through this standard definition, he wanted to promote an easier space access and hands-on university-level space education. Standardization of the bus allows standardization of the interface between the satellite and the launcher so that standard deployment systems can be developed and launch opportunities increased. As an educational university program, the CubeSat concept is based on lowering the costs and simplicity. The use of non-space rated Commercial-Off-TheShelf (COTS) components is thus unavoidable. The CubeSat leitmotiv is then :

In 2000, Pumpkin, Inc. decided to design a reliable off-the-shelf CubeSat Kit conform to CubeSat specifications to facilitate CubeSat development within tight schedules. As said in the presentation made Andrew E. Kalman, Pumpkin’s president, the CubeSat Kit is strong, modular, light, scalable, customizable and affordable [47]. Now, Pumpkin offers 1-unit (1U), 2-unit and 3-unit structures and also OBC, EPS and ADCS modules. In conjunction with Prof. Jordi Puig-Suari at California Polytechnic University San Luis Obispo, a Picosatellite Orbital Deployer (POD) has been developed : the P-POD (Poly-Picosatellite Orbital Deployer). It can launch up to 3 single-unit CubeSats or any fitting combination. Nevertheless, one had to waid until 2003 to see first CubeSats in space.
Lionel Jacques 2nd Master in Aerospace Engineering

6

University of Liège Applied Sciences Faculty Academic Year 2008 − 2009

Chapter 1.. the P-POD 1. As the lines of the magnetic field are already falling above Belgium.4 – Earth’s magnetic sure.. This explains the main goal of hysteretic materials: to prevent from too high rotation rates which could cause an additional unfavorable modulation of the communication signal.3 – On the left. on the right. Lionel Jacques 2nd Master in Aerospace Engineering 7 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 . even field orientation on the orbit[25] with hysteretic materials. OUFTI-1 project. This is favorable by seeing antennas’ radiation pattern.) are such as. the magnet will be perpendicular to the face with the antennas so that antennas are nearly perpendicular with zenith above Belgium. The initial rotation rate given at the P-POD ejection is variable. the CubeSat will never stop rotating. atmospheric drag. our CubeSat will be passively controlled with hysteretic materials and a permanent magnet. solar presFigure 1. But perturbation torques (gravity gradient.4. mission and nanosatellite Figure 1. On another hand. the permanent magnet will align the satellite on Earth’s magnetic field which is represented during one orbit on the figure 1. the CubeSat Kit Structure.4 The Subsystems ADCS [25] Since our payloads do not require a precise attitude control. Hysteretic materials will thus damp the rotational energy acquired at the deployment.

GND [10] The Ground subsystem consist in the tracking antennas.5 – OUFTI-1’s test RX PCB[36]. The EPS contains three power bus: 3. Through this. The first one can only deal with D-STAR transmission while the UTS also supports communication with the CW beacon. will be made of two separated units : the first one located at the Poste Central de Commande (PCC) of the university and the second one. Nevertheless. already presented in a previous paragraph. its remote control and the scheduler. Anyway. As previously presented. the remaining face being dedicated to the antenna deployment mechanism. As described previously. A Backup Tracking Station will : Telecommand. 5 faces of the CubeSat will be covered with 2 solar cells connected in series.3V. The exFigure 1.6 – Main EPS act battery model is not yet defined but engineering model Varta or Kokam lithium polymer batteries are nevertheless foreseen for their high energy density. In addition to the main and robust 3.25 protocol has been chosen for TC/TM1 to avoid TC/TM relying on the experimental communication payload. AX. TM : Telemetry states for Global Educational Network for Satellite Operations. mission and nanosatellite COM [36]. Concerning OUFTI-1.[33] The primary function of the EPS is to store the energy collected by the solar cells in the batteries and provide the bus with the required voltages. OUFTI-1 project. GENSO aims to increase the return from educational space missions.[28] Lionel Jacques 2nd Master in Aerospace Engineering 8 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . D-STAR communication protocol will be used as payload. GENSO is a worldwide network of ground stations and spacecraft interacting via a software standard. there is also a reliable CW beacon sending continuously satellite’s key housekeeping parameters that any hamradio operator in the world will be able to listen and forward to us. EPS [55]. located at the Montefiore Institute. 5V and 7.3V converter. the User Tracking Station (UTS).[28] OUFTI-1 will use the amateur radio communication bands : 435 MHz for the uplink and 145 MHz for the downlink. solar cells are our second payload and are provided AZUR SPACE.Chapter 1. 2 GENSO 1 TC Figure 1. The GENSO2 compatibility is thus foreseen for the UTS. located on the campus of the University of Liège. The Main Tracking Station.2V. an innovative digital EPS is being developed in cooperation with Thales Alenia Space ETCA.

antennas will have to be folded since the CubeSat will be in the P-POD until deployment.Chapter 1. they will be maintained by a small thread which will then be heated up to its melting point to release both antennas. The simulator include orbit propagation. COM for the link budget and Doppler considerations. Beukelaers in Simulink[9] Lionel Jacques 2nd Master in Aerospace Engineering 9 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 . OUFTI-1 will have two monopole λ 4 antennas : a 17cm and a 50cm one. link budget with the ground station calculation. To do this. Figure 1. ADCS for the attitude determination. During the launch phase. It will be fully redundant and linked by a high speed internet tunnel with the first one. Doppler estimation and so on. the MECH subsystem has to design a reliable deployment mechanism. A screenshot of the simulator is shown on the figure below. MIAS [9] Mission Analysis consists in analyzing the global comportment of the satellite in orbit. atmospheric drag. As Figure 1. OUFTI-1 project. . taking into account Earth’s oblateness.7. Therefore. solar radiation pressure. This includes lifetime evaluation. . A simple model of each subsystem is also embedded : EPS for power calculation. A thermal model will also be provided and described later.8 – Snapshot of the Simulator developed by V.7 – OUFTI-1’s antenna shown on the figure 1. MECH [63] Using two frequencies (uplink and downlink). . a global simulator is being developed in Matlab Simulink environment in correlation with STK software. mission and nanosatellite also be installed at the Redu Euro Space Center. . Throughout the launch sequence. antennas will be deployment mechanism panel[63] winded around a dedicated panel.

OUFTI-1 project. STRU [42] The first key objective of the Structure and Configuration subsystem is to position the many components in order to be compliant with the CubeSat design specification and many other constraints (thermal. OUFTI-1 will have two fully redundant OBC : a first robust one provided in the Pumpkin’s CubeSat Kit (FM430 flight module with Texas Instruments singlechip 16-bit MSP430). it will probably be used as the main and only OBC for the next OUFTI Missions.[19]. solar cells’ currents. Therefore. The picture 1. . The second objective is to ensure withstanding to the harsh launch environment : even in the P-POD.9 shows the home made OBC Printed Circuit Board (PCB). This is why a second lighter OBC has been developed (OBC2). OBC also deals with housekeeping parameters such as batteries’ voltage. . Those final tests will be performed at the Centre Spatial de Liège. deals with the different operational modes. the nanosatellite undergoes extremely strong vibrations.10 – Finite element random and sine vibrations tests are permodal analysis with formed in correlation with Samcef finite SamcefField[42] element models. But this one contains Figure 1. mission and nanosatellite OBC [54]. The CubeSat should ultimately proof that its structure and equipments are capable of withstanding the maximum expected launch environment through a series of acceptance and qualification tests. the idea to buy the CubeSat Kit structure and OBC module (FM430) was accepted to save time and let us concentrate on our payloads and other subsystems. Seeing the tight schedule. ensure telemetry data formatting and storage and telecommand data decoding and management. high frequency issues).[26] The On-Board Computer is the brain of the CubeSat : it controls data flow.Chapter 1. If this home proves itself. radiations. Lionel Jacques 2nd Master in Aerospace Engineering 10 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 . both Figure 1.9 – OUFTI-1’s second OBC PCB[19] many unnecessary functionalities for us. temperatures.

5 OUFTI-1’s anatomy Throughout this work. remove-before-flight pin hole and other ports are located.Chapter 1. The separation between PCBs is ensure by spaces which are small aluminum cylinders. All the PCBs are maintained together thanks to four endless screws. While the OBC is the CubeSat’s brain. The skeleton of the satellite consist of Pumpkin’s 1U structure itself divided into three parts : the lateral frame and the top and bottom faces. This PCB layout has been driven by many constraints and considerations among these the CubeSat specification stating that the center of gravity of the CubeSat has to remain inside a 4cm diameter sphere centered on the geometric center of the cube. Lionel Jacques 2nd Master in Aerospace Engineering 11 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . These components are therefore introduced here through relatively detailed exploded view (figure 1. The solar cells will be glued on aluminum panels. The remaining one is dedicated to the antenna deployment mechanism panel. its heart is the EPS with its two LiPo batteries fixed on a secondary PCB and is strategically positioned as the central PCB. From bottom to top. The last two are fixed on the main frame with M3 screws not represented here (4 screws for the up face and 6 for the bottom one). Inside OUFTI-1.11). As depicted on the view. The last PCB is the communication one on which will be located amplifiers. all aluminum panels will also be glued on the structure to avoid remachining. only five of the six faces of the cube will be covered with solar cells. one means the four vertical edges of the lateral frame which are hard anodized because they are the only surfaces in contact with the P-POD. By the rails. mission and nanosatellite 1. We called the ports or rear side of the CubeSat the side where USB. Above them lies the EPS also followed by the payload : the innovative EPS (EPS2). there are five printed circuit boards (PCBs). OUFTI-1 project. OUFTI-1 will indeed have two antennas described in a later paragraph. They are fixed at their bottom on the bottom face of the frame and at their top on the lateral frame with the "midplane standoffs". We already sees that there are only 8 contact points between the PCB stack and the external structure relies : four with the bottom face and 2 2 with the lateral frame. different components and specific parts of OUFTI-1 will be referred to.5mm). Up to now. one has the OBC (CubeSat Kit’s FM430 Flight Module) followed by its redundant home-made counterpart. quartz and other electronics circuits. the OBC2. The panels must be such as they keep the rails free (at least 8.

mission and nanosatellite Aluminium Panel Solar Cell Top aluminium frame Feet Midplane Standoff Batteries Battery PCB Spacer COM EPS2 EPS1 OBC2 OBC1 Lateral aluminium frame Antenna deployment mechanism panel Ports side 𝑧 Bottom aluminium frame 𝑥 𝑦 Figure 1.Chapter 1.11 – OUFTI-1 exploded view Lionel Jacques 2nd Master in Aerospace Engineering 12 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . OUFTI-1 project.

The Thermal Control Subsystem

2

This chapter introduces the basics of spacecraft thermal control. It begins with a brief recall of heat transfer and then thermal environment will be exposed. After that, we will describe in a few words what is the aim of the thermal control subsystem : why and how dealing with space thermal issues.

2.1

Space heat transfer
In general, there are three main heat transfer modes1 : conduction, radiation and convection.

2.1.1 Conduction
In space, due to the extremely low residual pressure, only conduction and radiation modes are present2 . Conduction is governed by Fourier’s Law. For an isotropic material: q= krT

where k is the constant thermal conductivity and where q is the heat flux. This equation can be rewritten in the case of a steady unilateral flow through a surface of a thickness L and constant area S as follows : kS ∆T = GL(i,j) ( Ti Tj ) L This defines the thermal conductance GL(i,j) between the isothermal surfaces i and j. Q = qS =

2.1.2 Radiation
On the other hand, radiation heat transfer is governed by StefanBoltzmann’s Law stating that the black-body irradiance is proportional to the fourth power of its temperature :
1 Ablation, a combination of these three processes with chemical reaction, is also considered for atmospheric re-entry vehicles 2 Nevertheless, convection must be taken into account for manned mission, launchers. . . (ISS, Shuttle,. . . )

13

Chapter 2. The Thermal Control Subsystem

EBB ( T ) =

∞ 0

EBB (λ, T )dλ = σ0 T 4 [W/m2 ]

where σ0 is the Stefan-Boltzmann’s constant in vacuum defined in function of the universal constants : π, Boltzmann’s constant k, the speed of light in vacuum c0 and Planck’s constant h : 2π 2 k4 = 5.67051 10 8 [Wm 2 K 4 ] 15c2 h3 EBB (λ, T ) is the hemispherical spectral emissive power of a black-body and is given by Planck’s Law : σ= EBB (λ, T ) = 2πhc2 λ5 (e kλT
hc

1)

[W/(m2 .µm)]

Figure 2.1 – Planck’s black body radiation curves for different temperatures

Planck’s law is illustrated in the figure 2.1. For a given temperature, there is a maximum in the energy distribution. The corresponding wavelength is governed by the Wien’s displacement law describing the location of the maximum in the hemispherical spectral emissive power : λmax = b [m] T

with b = 2.897 10 3 [m K ]. All these formulas concerned the black-body which is idealized object absorbing all radiant energy from any direction or wavelength and emitting in any direction isotropically. The radiated energy of the black-body only depends on its temperature. But a real body can absorb, reflect and transmit radiation energy so that absorptivity α, transmittivity τ and reflectivity ρ quantities are defined, all wavelength and angular dependent. For a semi-transparent body, energy conservation leads to the following equality α(θ, λ) + ρ(θ, λ) + τ (θ, λ) = 1

Lionel Jacques 2nd Master in Aerospace Engineering

14

University of Liège Applied Sciences Faculty Academic Year 2008 2009

Chapter 2. The Thermal Control Subsystem

where θ is the angle of incidence. As there is no perfect black-body in practice, the emissivity (λ) is defined as the ratio between the energy emitted by a surface to that of a black body at the same temperature. Absorptivity and emissivity can either be hemispherical or directional and either total or spectral. The second Kirchhoff’s law states that for a given direction θ, directional spectral absorptivity and emissivity are equal :

(θ, λ) = α(θ, λ)
But, in general, this is not true with total hemispherical values mainly because of their strong wavelength dependence. Both α ans varies with the angle of incidence but they are assumed to follow the Lambert’s law stating that directional absorptivity/emissivity is proportional to cos θ (maximum for normal incident angles and null for tangential ones). The reflectivity can be either diffuse or specular (ρ = ρd + ρs ). When diffuse, it follows the Lambert’s law as the emissivity and absorptivity while specular reflection follows Descartes’ law according to which the reflexion angle is equal the the incident angle. In space heat transfer, thermal engineers made the following assumption : thermo-optical properties are assumed to be constant in two spectral regions : - the infrared spectrum, from λ = 4.25µm to 40µm, corresponding to temperatures between 70K and 700K. - the visible spectrum, ranging from λ = 0.3µm to 2.5µm associated to a temperature range going from 1150K to 10000K As the temperature of a spacecraft lies in the 70K-700K range, the emitted radiation is infrared. But the source of the main incident radiation is the sun which can be considered as a blackbody emitting at 5776K. This temperature lies in the visible spectrum, as it is also shown by the figure 2.1. Actually, ESA and NASA thermal engineers adopted the following convention : they call the (constant) emissivity (and absorptivity) in infrared wavelengths and α the (constant) absorptivity (and emissivity) in visible wavelengths. This convention will also be adopted in this work. Finally, surface thermo-optical properties are defined through eight coefficients (6 independent) related by the two following equations :

+ ρ IR,d + ρ IR,s + τIR = 1

α + ρV IS,d + ρV IS,s + τV IS = 1

Where τ is the transitivity An important concept for radiative exchange is the view factor F(i,j) between two surfaces Si and S j . It is the proportion of all the radiative power leaving Si which directly strikes surface S j (without any reflection) and only depends on the geometrical configuration : Fi,j := Pij 1 = Pi πSi cos θi cos θ j
Si Sj 2 rij

dSi dS j

Another useful concept is the Gebhart factor Bij : this factor takes into account multiple reflections and represents the portion of the radiation emitted by a surface Si and finally incoming to the surface S j . The Gebhart factor does not only depends on geometry but also on thermo-optical
Lionel Jacques 2nd Master in Aerospace Engineering

15

University of Liège Applied Sciences Faculty Academic Year 2008 2009

the galaxies and the Cosmic Microwave Background which will be studied by ESA’s new satellite Planck. only two heat sources are still present : Earth’s infrared and internal dissipation and the spacecraft will be cooler.j) σ( Ti4 Tj4 ) This defines the radiative exchange factor GR(i. 2. At equilibrium. The sky. valid for diffuse reflection1 : Bij = Fij j + ∑ Fik (1 k k ) Bkj The radiative coupling between two surfaces is directly deduced from the Gebhart factor. 2. The Thermal Control Subsystem properties. 4. During the eclipse. For Earth. the mean reflectivity is assumed to be near 30%. Earth infrared radiation. is the main source of cold and can be seen as a black body emitting at 3K. Internal dissipated power in electronic components (Joule effect). with a mean value around 1367 [W/m2 ] at 1AU (1414 [W/m2 ] at winter solstice and 1322 [W/m2 ] at summer solstice [24]). Albedo planetary reflected radiation. aerothermal flux have also to be considered during launch or re-entry phases.2 The thermal environment A satellite orbiting Earth has several heat sources. This temperature represent the radiation of the stars. Earth can be modeled as an equivalent black-body emitting at 255 K [50]. Out of equilibrium : Qij = 4 i Si Bij σTi 4 j S j B ji σTi = 4 i Si Bij σ ( Ti Tj4 ) = GR(i. Not represented on the figure. Indeed. It is defined through the following equation. the radiative power Qij flowing from Si to S j is equal to the difference between the power emitted by Si . Direct solar flux depending on sun distance. The temperatures of the satellite tend thus to vary in a cyclic way along the orbit. Qij = 0 and thus i Si Bij = j S j Bji . 5.j) between two surfaces i and j. rising in sunshine and dropping during eclipse. But it can vary localy up to 40 or 80% above shiny clouds and from 5 to 10% for ocean and forests[44].2 : 1. theory has been developed to take partial specular reflection into account [50] but will not be used in this work because the reflections encountered here are mainly diffuse 1A Lionel Jacques 2nd Master in Aerospace Engineering 16 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . 3. as represented on the figure 2. [24]. [44]. called Deep Space. absorbed by S j ( i Si Bij σTi4 ) and the one emitted by S j and absorbed by Si ( j S j Bji σTi4 ).Chapter 2.

there are two main categories of thermal control means : active and passive ones.MLI blanket . that will be later introduced for OUFTI-1.2 – Thermal environment of a satellite in LEO 2.Phase change materials Heaters . stand-off Passive Heat pipes & fluid loops .1 – Thermal control means Lionel Jacques 2nd Master in Aerospace Engineering 17 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 . The Thermal Control Subsystem Direct Solar Flux Internal dissipated power Albedo Earth Infrared Emitted radiation Figure 2. filler. Then.loop heat pipe . For precise pointing satellites.3 Thus.mono/diphasic fluid Louvres Active Table 2.coating . Therefore.washer.Chapter 2. why Thermal Control ? and how ? The primary objective of thermal control is thus to maintain temperatures of onboard equipments within specified ranges.doubler. the critical equipments usually are optical instruments and electronics components (mainly the batteries).structural materials . strap. how to fulfill all these requirements since the spacecraft undergoes temperature variation solely determined by its radiative exchanges with environment.fixed/variable conductance . It also must ensure that temperature gradients (spatial consideration) are not too large and a good temperature stability (temporal consideration). Temperature gradients are undesirable for optics pointing and alignment while stability and thus narrow temperature ranges are important for the sensitivity of detectors.ground control Peltier element Conduction . bolt.thermostat control . In fact. This can be summarized in the following non exhaustive table : Radiation . adhesive .Thermal protection system . thermo-elastic coupling involving unwanted vibrations also occurs at eclipse in and out : this is known as jitter phenomenon.electronic control .radiator Latent heat & Ablation .

Chapter 2. The Thermal Control Subsystem

Except for heaters, most of active means required too much power and mass or complex mechanisms and moving parts or even fluid packaging and this is not compatible with the CubeSat concept. Furthermore, even if active thermal control means are able to cope with larger heat loads, they are also generally less reliable than passive ones and CubeSats have usually low power consumption and dissipation. Nevertheless, if the temperature of an equipment dropped under its specified minimal one, a simple way of heating is the use of heaters. The advantage of heater is that it can nearly be of any dimension and is thus very convenient. For all these reasons, the main and easiest way of doing thermal control for nanosatellites remains passive control or small heaters.

2.4

Thermal control systems of other CubeSats
This is indeed the means other CubeSat teams have used to do thermal control. Some of them are described below: SwissCube SwissCube nanosatellite [53] is developed at EPFL in Lausanne, Switzerland. SwissCube’s mission is to study the nightglow phenomon. SwissCube has its own developed structure fabricated by CNC milling and wire electrical discharge machining and uses an active attitude control with magnetotorquers, sun sensors, gyroscopes and magnetometers to stabilize the satellite. Concerning the thermal control subsystem, their simulations led them to use a 500 mW (25 Ω) heater to keep the two batteries within their allowable range of temperatures. They are packed in an aluminum box to prevent deformation due to the vacuum environment and are fastened on a copper plate as the heater, all with epoxy resin. The copper plate is used because of its high thermal conductivity to conduct the generated heat to the batteries. The heater is thermostatically controlled avoiding the use of the OBC and is switched on under the 0 C threshold. SwissCube has not yet been launched but is ready for it. Compass-1 Compass-1 [12] is developed by the students of the University of Applied Sciences in Aachen. Its mission is to take picture of the Earth and it has therefore a active attitude control as SwissCube. Their thermal model led them to use 2.01 W (68 Ω) heater located at the top of the battery box combined with LM75 temperature sensors. The heater is switched on once the temperature of the batteries drops under 5 C. Compass-1 has already been launched on April, 28th , 2008 by the Indian PSLV launcher. After the successful launch, Compass-1 entered in a vicious circle: the heater was too often switched on and the battery power drops drastically. This untimely switch on of the heater occurred because the temperature to which was compared the threshold was based on the minimal temperature given by the three sensor inside the battery box. The threshold has then been adapted but Compass-1 had to be reset for other reasons so that the nominal threshold had also been reset. Unfortunately,
Lionel Jacques 2nd Master in Aerospace Engineering

18

University of Liège Applied Sciences Faculty Academic Year 2008 2009

Chapter 2. The Thermal Control Subsystem

they lost their two ground stations at that time but, with the help of the amateur radio community, they managed to recover the satellite and successfully changed into its nominal mode. Now, all goes well for Compass-1 and many pictures have already been taken. Del -C3 Delfi-C3 [13] is the nanosatellite developed and has been launched with Compass-1 in April 2008. In addition to be a 3-Unit CubeSat, on the contrary of both previous CubeSats, the particularity of Delfi-C3 lies in the fact that it do not have any battery but deployable solar arrays. As the battery is usually the more critical component, Delfi-C3 has no active thermal control and its thermal behavior relies only the thermo-optical properties of the outer surfaces.

Lionel Jacques 2nd Master in Aerospace Engineering

19

University of Liège Applied Sciences Faculty Academic Year 2008 2009

Preliminary thermal analysis

3

A first analysis had already been developed in the Phase A study by S. Galli. That was a steady state model and it did not take into account the effect of satellite’s thermal inertia. In this chapter, a single node transient model of the CubeSat will be implemented in Matlab Simulink environment. This chapter will begin with a description of the notations and conventions that will be adopted not only in this preliminary analysis but also in the next chapters. Then, we will study the thermal environment and implement a general way to compute the different incoming fluxes. After that, the cyclic single node transient model will be introduced and first guesses of hot and cold case will be computed. A sensitivity analysis will also be performed.

3.1

Notations
Here are described the notations and conventions we will use throughout this work. Concerning the reference frame, it will be the one used by the Structure and Configuration subsystem. It was shown on the exploded view1 (figure 1.11) : its center is the geometric center of the CubeSat. The + X direction will be perpendicular to the face where are fixed the antennas and where the access ports are located and oriented to the opposite face, the + Z direction will be the direction perpendicular to PCBs stack, from bottom with the OBC PCB to top with the COM PCB. Finally, the +Y direction is such as it forms with the other ones a right handed coordinate system. We will also use the following numbering convention :

+X = 1 & +Y = 2 & +Z = 3 &

X=4 Y=5 Z=6

Concerning thermo-optical properties, ESA and NASA convention will also be adopted, as described previously : α = αvis =
1 Excepted

vis

that it is not located at the center of the CubeSat for the sake of clarity

20

. from 0 to +45 C. and also for the PIC18F2331 microcontroller used on the experimental EPS [37]. To avoid this potential problem. In this case. and minimize the difference between the thermal expansion coefficients of the Lionel Jacques 2nd Master in Aerospace Engineering 21 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . all have the same specifications depending on whether it is in charge. A more critical component was the oscillator because of its important temperature sensitivity. Typical values for solar cells operating temperature ranges are -100 to +100 C [62]. the ADF7021 transceiver has also the same wide temperature range [5]. it was decided. the most critical part of the satellite remains the batteries. following its datasheet [48].2 Thermal Requirements As explained in the previous chapter.Chapter 3.1. Concerning COM PCB integrated circuits. this is the case for Pumpkin’s FM430 Flight Module (OBC). the use of a TCXO (Temperature Compensated Quartz Oscillator) has been foreseen. Finally. The main CubeSat Kit’s structure is made of aluminum 5052H32 alloy and thus already satisfies the differential dilatation condition with the P-POD. among the possible foreTM seen models (VARTA PoLiFlex R [58] and Kokam SLPB 554374H [31]). common TCXOs have the same temperature range. also made of aluminum alloy (7075-T73). Lithium polymer batteries have been chosen for their high specific capacity. in agreement with the Structure and Configuration subsystem. Panels will indeed be glued on the main frame (for a reason already explained in a previous chapter). the aim of thermal control is to ensure that all onboard units will be working within their own allowable temperature range. The CubeSat Kit’s. Component Main structure Solar cells Electronics LiPo Batteries Tmin [ C] -40 -100 -40 0 -20 Tmax [ C] +85 +100 +85 45 60 Note charge discharge Table 3. Preliminary thermal analysis = IR = α IR Environmental fluxes notation will be as follows : QS Solar power Q A Albedo power Q E Earth infrared power Q I Internal dissipated power 3.1 – OUFTI-1’s Thermal Requirements Another thermal requirement concerns the thermal expansion of the structure. All electronic components are not yet defined but the use COTS components mainly involves -40. Concerning the panels supporting the solar cells. from -20 to +60 C. also operates in the -40 C to +85 C industrial temperatures range [47]. OUFTI-1 thermal requirements are summarized in the table 3. For instance. The final model is not yet determined but. as all the previous parts. . +85 C temperature ranges. that aluminum would again be more suitable than CFRP for the same reason. or discharge.

1) Lionel Jacques 2nd Master in Aerospace Engineering 22 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . 3. the different incident fluxes exposed in the previous chapter will be calculated.3 Thermal environment related to OUFTI-1’s orbit Now.1 The orbit An orbit is described by its classical orbital elements (COE aka Keplerian elements) : the semi-major axis a.1 – Classical orbital elements The reference line for the RAAN is taken as the vernal equinox. To compute the satellite position around the earth. a simple Simulink Model will be developed in order to integrate the equations of motions of the two-body problem and compute the position of the satellite required for the fluxes calculation.1: a= µ kr k3 r (3. To do so. the right ascension of the ascending node Ω or RAAN.Chapter 3. the intersection of the equatorial plane with the ecliptic at vernal equinox pointing directly to the sun. i.3. the inclination i.e. Preliminary thermal analysis panels and the frame was a good way to reduce the shear stress inside the glue. the RAAN and perigee argument ω. OUFTI-1’s orbit has a perigee altitude of 354km. an apogee altitude of 1447km and an inclination of 71 . a simple Simulink two body model has been derived. the eccentricity e. Initial position and velocity computed from the COE are given as input to the Simulink 6DoF Euler Angles integrator in addition to the acceleration given by equation 3. are still unknown. the argument of perigee ω and the true anomaly ν. 3. Figure 3. The two remaining parameters.

Alet and an abacus coming from the book "Spacecraft Thermal Control Handbook" by D. i. This is done in the Albedo and IR calculation box described in the next paragraphs. it is the rotation matrix to pass from the inertial coordinate frame to the body axes. The rotation rates of the satellite will remain constant as we consider that no moment act on it. Then.e.3 Albedo flux The albedo flux is relatively more complex to evaluate. any ray that is reflected by one face will never hit another face. described in [56]. we decided to combine two references : a simple formula coming from the notes of "Contrôle thermique des engins spatiaux"[1] by I. The sun direction will be taken as a constant and equal to rsun = [1 0 0]. 3. the integrator gives the position and direct cosine matrix (DCM) for the next time step. at each time step. the rays coming from the sun are assumed to be parallel. For that reason. Many different expressions or tables can be found in the literature to compute the albedo but they are often not easy to use or incomplete. Gilmore [24]. This method is based on determining if the line between the satellite and the sun intersects the Earth : if it does. This means that if the inclination is 90 . The advantage of this method is that it avoids trigonometric operations. the orbital plane will be perpendicular to the sun rays if Ω = 90 and parallel for Ω = 0 .i = CS (ni rsun ) [W/m2 ] The incident solar flux on each face is thus directly determined by the direct cosine matrix (DCM) and only depends on the solar constant CS and the satellite attitude. refer to [56] and [9].Chapter 3. there is no coupling between the faces.e. Therefore.3 : 1 initially coming from Lockheed Martin Lionel Jacques 2nd Master in Aerospace Engineering 23 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . But how to know whether the spacecraft is in sunlight? The same method as the one implemented by V. seeing the cubic shape of the satellite. For further details. Earth view factor and many others. i. Furthermore. the incident flux on each face is simply obtained by multiplying the solar constant CS with the scalar product between the normal to the face and the sun direction. An approximate attitude is given by the mean of initial euler angles and rotation rate of the satellite. the model compute the incident heat flux on each face of the satellite. Here is the formula and the abacus1 is represented on the figure 3. developed by Alfano in 1991. 3. Beukelaers in the Simulator was used: the Line-Of-Sight (LOS) method.3. As output. there is no LOS and the satellite is in eclipse and in all other cases it is in sunlight. Indeed. Preliminary thermal analysis where µ is the standard gravitational parameter of the earth and r the position vector of the satellite.3.2 Solar flux As the sun distance is extremely large. it depends on many parameters such as satellite’s position from the subsolar point. qS. The DCM represents the current orientation of the satellite compared to the inertial axes.

2 – Simulink two-body propagator for incident ux computation : main window (top) and Albedo and IR calculation box (bottom) Lionel Jacques 2nd Master in Aerospace Engineering 24 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .Chapter 3. Preliminary thermal analysis x_ECI a_ECI Matrix Multiply a − C− DCM r_sun _ECI albedo IR 1 2 3 4 5 Gravitationnal Model r0 r_sat _ECI albedo _face Matrix Multiply 6 7 8 9 10 11 a r0 v0 4 Cs IR _face DCM Solar _face Normal 6 7 8 9 10 11 e nu Raan omega i pqr0 Xe 5 euler0 v0 a Albedo and IR calculation Equations of Motion Integration 12 13 14 Initial Conditions 1 p0 2 q0 3 r0 : COE 15 16 17 18 19 Epoch 20 U( : ) Reshape 21 12 phi 13 psi 14 omega 3 Cs Cs Albedo 1 albedo 5 a a IR 2 IR sqrt Dot Product u− 6378 Bias Altitude alb _IR Albedo_face rhos 3 albedo _face 2 r_sat_ECI 1 r_sun_ECI Normalize r_sat fcn theta Normalize r_sun theta IR _face 4 IR_face Earth Infrared and Albedo computation r_sat fcn rhos 4 DCM DCM r_sun_ECI cosines fcn DCM Normals 3_ECI 6 Normal 5 Solar _face Sun_ECI fcn sunlight Sat_ECI Figure 3.

The 0.3 – Incident albedo irradiation on a surface element in Earth orbit [24] The formula required to compute the view factor between the face and the earth.Chapter 3. even if this view factor can be easily hand computed with the abacus. θ the "sun-earth-satellite" angle and FE the view factor between the face and the earth. When θ = 0. Therefore.1 sin Lionel Jacques 2nd Master in Aerospace Engineering ρ 2 e (3. the satellite is at the subsolar point and the albedo is maximum. But. Figure 3. to simplify their use. Preliminary thermal analysis q A = CS a [cos(0. all these curves were digitized and. a the albedo reflectivity coefficient. this is no more easy to implement within Matlab. reflected rays are still hitting the satellite. fitted with this formula : FE = r2.5 FE [W/m2 ] where CS is the solar constant. FE which is quite difficult since it depends both on the altitude and orientation of the face.9 coefficient in the cosine means that even if the satellite has just come above the shadow part of the Earth.2) University of Liège Applied Sciences Faculty Academic Year 2008 2009 25 .9 θ )] 1.

4 Earth infrared flux Earth’s infrared flux.2 The equation and the abacus show that even if the scalar product n f ace rsat is negative (i.3.2 10000 0. one can obtain the following equation : q E = EBB j T =255K FE = 240 FE [W/m2 ] Lionel Jacques 2nd Master in Aerospace Engineering 26 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .7 View factor FE 0. 1 Lockheed Martin’s Data Approximation 0.31r6 + 723.e. This albedo model does not take into account local reflectivity variation (clouds. This can be explained by the fact that the incoming rays are no more parallel as it was the case for the sun rays since the satellite is close to the emitting body that can no more be considered as a point. ρ < 90 deg and the face does not point directly toward the Earth).6r3 780. in comparison with the Lockheed Martin abacus.232 where r = RR+h . Preliminary thermal analysis with the exponent e described as a function of r: e= 160. also depending on the same FE is now easier to compute since FE is already known.6 2000 0.4 – Fitting of the Gilmore abacus curves with the formula 3.5 3000 0.8 1000 0.3 6000 0.4. forests.65r2 + 226. . This feature will involve a greater effective surface as it will be shown.9 Altitude [km] 200 500 0.1 20000 0 0 20 40 60 80 100 120 140 160 180 ρ [deg] Figure 3. The fitting approximation of the view factor FE is represented on the figure 3. the combined model has been correlated with the abacus method. By neglecting diurnal and seasonal variation in the temperatures of the earth and considering it as an equivalent 255K black body (thus emitting diffusely in all directions). the view factor is not null and the face still absorbs albedo or infrared flux.36r5 1380r4 + 1394. .Chapter 3. oceans. Of course.4 4000 0. ) 3.81r 21.

400 Albedo Earth infrared 350 300 Incident fluxes [W/m²] 250 200 150 100 50 0 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 3. unless a face points perfectly away from the Earth. the aerothermal flux decrease with the altitude from 1300W/m2 at 150km to 160 then 39W/m2 at 200 and 250 km.5 – Nominal Earth IR and Albedo uxes seen by a surface pointing to the Earth on OUFTI-1’s orbit But it has already been noticed that when the satellite is close to the earth. Ω = 0 (line of nodes perpendicular to the sun vector). all the six faces have a non zero view factor FE . on OUFTI-1’s orbit.6 shows this. 3. 1 qnom n f aces Ae f f = ∑ i Ai qi = A qnom n f aces ∑ i Ae f f 1 qi ) = A qnom n f aces ∑ i qi The figure 3.3. This contribution will then be neglected throughout this work. The CubeSat is initially oriented in such a way that its great diagonal is parallel to the local zenith. This means that an effective area can be defined as the area pointing toward the Earth that would receive the same flux as the sum of all the faces.Chapter 3.5 Aerothermal flux For a circular orbit. In Figure 3. Preliminary thermal analysis 3. ω = 0 .3. For this position. The remaining aerothermal flux at 350 km is only 7W/m2 . we are now able to represent the fluxes seen by the satellite on its orbit. all the six faces have a non-zero incident flux and that the effective area ratio (ratio between the effective area and a cube’s face) is greater than 2 ! Lionel Jacques 2nd Master in Aerospace Engineering 27 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .5 is plotted the nominal Earth IR and albedo fluxes incident to a unit surface pointing toward the Earth (ρ = 180 ).6 Results Using this model.

Preliminary thermal analysis 200 Incident infrared flux [W/m²] 150 100 50 Face 1 Face 2 Face 3 Face 4 Face 5 Face 6 0 0 10 20 30 40 50 60 Time [min] 70 80 90 100 2. The effective area ratio does not significantly change and this is the same for albedo and IR. The conclusion is that the spin rate of the CubeSat does not significantly affect the effective area ratio related to IR and Albedo for LEO. corresponding to the case when the Lionel Jacques 2nd Master in Aerospace Engineering 28 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .5 2 1.5 0 Effective area ratio 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 3.7.5 1 0.7 – Effective area ratio On another hand.5 0 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 3.5 Solar IR Albedo 2 Effective area ratio 1.Chapter 3.5 1 0. 2.6 – Infrared uxes incident on each face and effective area ratio The case of a randomly spinning CubeSat is represented in Figure 3. the effective area ratio related the solar flux never p exceeds the theoretical value of 3.

j are respectively the conductive and radiative links between nodes i and j.i ) The cyclic transient solver is implemented in Matlab/Simulink environment. Preliminary thermal analysis great diagonal is pointing to the sun (the projected area is than the cube’s face area).1 Implementation The basic underlying equation used for this model is the lumped parameter equation. This equation results from the heat balance to node i.j ( Ti Tj ) + σ j=1. Lionel Jacques 2nd Master in Aerospace Engineering 29 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . For a single node model. Ti is the temperature of the node i and Ci its heat capacity. For this preliminary analysis. The solver is made up two main consecutive phases : a first one calculating the external incident fluxes around the orbit in Simulink (already developed and represented at the figure 3. Qexti the total external power incoming on the node i (Qexti = QSi + QAi + QEi ). By this way. GRsat.j6=i ∑ n GLi. after a finite number of orbit integrations. the temperature must be computed.j and GRi.j ( Ti4 Tj4 ) = Ci dTi dt (3.3) Qinti the total internal dissipated power for the node i. GLi.i + i q E.i + αi q A.DS is simply equal to the product of the CubeSat’s total area by its averaged emissivity : 4 Qint + Qext + σGRsat. As there is no coupling between the faces of the cube. p 3 times greater 3.3 is integrated over one orbit several times. it reduces to : dTsat dt where GRsat. written here under its general form for a transient multiple nodes model : Qinti + Qexti + j=1. It is based on only one node for all the satellite (so considered isothermal) but taking transient effects into account. Once the first phase is completed and the external absorbed power is known for each face and at each position around the orbit. a very simple model has been created. 3.2) and a second one integrating the lumped parameter equation.Chapter 3.DS is the radiative exchange factor between the satellite and the Deep Space and TDS = 3K the Deep Space temperature. the equation 3.j6=i ∑ n GRi.DS ( Tsat 4 TDS ) = C eq = ∑i Ai i ∑i Ai Concerning Qext. the heat flux incident on face i are simply multiplied by its corresponding equivalent absorptivity/emissivity (α for solar and albedo and for IR) and then added: n f aces Qext = QS + QA + QE = ∑ i Ai (αi qS.4 Cyclic Transient Model Now that the incident fluxes on each face can be computed.4.

3. the Runge-Kutta algorithm (ode45 in Matlab) is used to solve the non linear differential equation over the orbit period.2 Validation Before running and analyzing the model.8 – Cyclic transient Matlab Simulink solver ow chart The cyclic repetition of the temperature evolution is based on two convergence criteria.Chapter 3. The time derivative of the temperature at the beginning and at the end of one orbit must be equal. Equations of motion integration with Simulink during one orbit Solar. as described in the flow chart. Ttk=0 Ttk=Torbit < tolT where Torbit is the period of the orbit. 2. The temperature reached at the end of one orbit must be equal to the temperature at the beginning of this orbit. The process is summed up in the flow chart of the figure 3.8. the initial temperature of the next orbit is set as the final one of the previous orbit. If they are not equal. the initial condition of the (k + 1)th orbit integration is updated with the final temperature of the previous orbit. If the convergence criteria are not satisfied. dT k dt dT k dt < toldT t= Torbit t =0 In practice. Preliminary thermal analysis the temperature evolution over one orbit does not depend anymore on the first initial condition. it has been correlated with ESATAN/ESARAD softwares. Albedo and IR incident fluxes computation Total effective absorbed power on one orbit Set initial 𝑇0 Thermal transient equation integration during one orbit 𝑘 𝑘+1 Set 𝑇0 = 𝑇𝑒𝑛𝑑 No Converge ? Yes 𝑇(𝑡) Figure 3. The validation is performed for a nonrotating cube pointing toward the sun on OUFTI-1’s orbit (354 1447km Lionel Jacques 2nd Master in Aerospace Engineering 30 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . 1.4.

CES EduPack. containing an impressive database of materials and process properties. Preliminary thermal analysis 71 ). 0 −5 Temperature [°C] −10 −15 −20 −25 −30 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Matlab/Simulink ESATAN/ESARAD 15 Matlab/Simulink Sun ESATAN/ESARAD Sun Matlab/Simulink Albedo ESATAN/ESARAD Albedo Matlab/Simulink Infrared ESATAN/ESARAD Infrared Power [W] 10 5 0 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 3. with a RAAN Ω = 45 . and a perigee argument ω = 0 . the heat capacities of OUFTI-1’s main components1 have been added. The cube is considered as a perfect black-body (α = = 1).6 32.3 776 Table 3.9 – Validation of the cyclic transient Matlab/Simulink solver with ESATAN/ESARAD 3.2 – OUFTI-1s thermal heat capacity The figure 3. created by Professor Mike Ashby of Cambridge University.9. it is time to briefly 1 All materials properties data comes from CES EduPack.Chapter 3. shows that both temperature and flux are identical with the two solvers and that the Matlab/Simulink one is valid.3 42.1 161. Component aluminum frame aluminum panels PCBs Antenna panel Batteries Total Mass [g] 145 165 350 33 50 698 Specific Heat [J/kgK] 980 980 1136 980 960 Heat capacity [J/K] 142. The different contribution are exposed in the table 3. Lionel Jacques 2nd Master in Aerospace Engineering 31 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .2. To evaluate its heat capacity.7 397.5 Results Now that one can convincingly compute the mean temperature of the CubeSat for given thermo-optical properties and orbit. is a very useful tool for materials selection.

3: Parameter Orbital parameters Solar constant Albedo coefficient Earth temperature Internal dissipation Hot Case permanently illuminated 1414 [W/m2 ] 0. A sensitivity analysis will be performed to see the relative importance of each thermo-optical property. But at this point. Albedo. as these values are not yet perfectly knows and defined. considering the solar constant at the winter solstice (1414 [W/m2 ]) instead of the mean value 1367 [W/m2 ] or a higher albedo coefficient or even a particular orbit (as the orbit changes over time). Preliminary thermal analysis introduce the concept of cold and hot case.35 250K (. The cold and hot case assumptions are summarized in the following table 3. For instance. as it will be reused many times throughout this work. On the contrary. These are indeed the worst cases. It is based on a worst case conservative study : the hot case corresponds the case for which all parameters. it is totally converted into heat in the hot case (cells’ efficiency equal to zero). the cold case is the perfect opposite of the hot case: all parameters are such as the reached temperature is minimal. On the other hand.3 – Preliminary Hot & Cold case definition As described in the table.Chapter 3. 220[W/m2 ]) full Cold Case max eclipse time 1322 [W/m2 ] 0. As the emissivity does not suffer great variation over time while the absorptivity tends to increase. The concept of Beginning of Life (BOL) and End of Life (EOL) values is then useful. all the electrical power collected by the solar cells is assumed to be perfectly stocked in the batteries without any dissipation. IR) Spacecraft Orbital Orientation Internal Power Dissipation External surfaces emissivity/absorptivity values are not perfectly known and vary throughout the lifetime of the spacecraft (mainly an increase for α due to high reactive atomic oxygen present in the upper layers of the atmosphere) and are therefore subjected to the cold/hot case definition. The parameters that are biased hot or cold are: Emissivity and Absorptivity Environmental Constants (Solar. One finally considers that the satellite is randomly rotating for both cases1 . 260[W/m2 ]) none Table 3. for the cold case. simulations [25] shows that the satellite will never stop rotating because of the passive control and continuous perturbation torques acting on it 1 ADCS Lionel Jacques 2nd Master in Aerospace Engineering 32 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . this effect will not be taken into account. both environmental and satellite related.25 260K (. BOL values are considered for the cold case analysis and EOL for the hot case. are chosen in such a way that they all contribute to reach maximal temperatures and/or gradients during one orbit.

The temperature falls down to -20 C. the lower the temperature reached. the mean temperature of the CubeSat in this preliminary hot case is already close to it and will thus be carefully looked at in the next analysis. 3. The figure 3.1 Hot Case As explained above. defined by the RAAN. But in this situation.Chapter 3. the first hypothesis used for the hot case is a constantly illuminated orbit which is possible due to the high inclination of OUFTI-1’s orbit : 71 . This would represent the mean temperature of the CubeSat and one can imagine that external parts could still be cooler and internal parts slightly hotter.10 – Hot Case definition. One could wonder if an orbit containing the subsolar point with a minimum eclipse time (perigee in eclipse) could be more critical. Recalling the maximal specified temperature of the batteries of 45 C. Preliminary thermal analysis 3. in uence of orbit orientation One can see that the lower the angle between the line of nodes and the sun direction. Figure 3. is perpendicular to the direction of the sun. This case would be slightly worst than the presupposed one but the difference is small and the hot case will still consider an orbit whose line of nodes is perpendicular to the sun’s direction.5.10 shows the influence of the orbit orientation on the hot case temperatures evolution. But there’s an angle near 70 for which the satellite still experiences no eclipse and yet have a slightly higher albedo.2 Cold Case Here are the results of the cold case defined earlier. Lionel Jacques 2nd Master in Aerospace Engineering 33 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . This means that the line of nodes. albedo is reduced because the satellite is continuously far from the subsolar point.5. When the satellite comes out of eclipse. the batteries begins to charge and this analysis let us think that the batteries could be below their minimal specified temperature while charging: 0 C.

For instance.11 – Cold Case temperature evolution 3. the figure 3.12 represents the evolution of Tmin & Tmax for all combination α/ of the aluminum panels. The remaining 20% are divided up between the frame and antennas’ panel. To do so.9 9. we carried out an evaluation of the sensitivity of the extremal temperatures to thermo-optical properties.6 Sensitivity analysis Except for the solar cells. we vary the α/ values for each component separately and compute the minimal/maximal temperatures in the cold/hot case.2 68 65.6 57. Figure 3.4 – External area ratios The solar cells obviously fill the half of the total area of the CubeSat while the aluminum panels and rails cover 30% of the cube.Chapter 3.4. Linearized sensitivities of the maximal and minimal temperatures of the CubeSat to the emissivity and absorptivity of the aluminum panel can thus be computed.6 100 Table 3.8 107.3 10. Component Solar cells aluminum panels Anodized rails Remaining aluminum frame Antenna deployment mechanism panel Total area Area [cm2 ] 301. One can already compute the area ratios of the different external materials of the CubeSat. Preliminary thermal analysis 10 5 0 Temperature [°C] −5 −10 −15 −20 0 10 20 30 40 50 60 Time [deg] 70 80 90 100 Figure 3.9 11. absorptivity/emissivity values are not yet well known. The coefficients for the other surfaces are obtained by the same way: all Lionel Jacques 2nd Master in Aerospace Engineering 34 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .3 17.4 600 Ratio [%] 50.12 shows that all the curves are nearly linear. That’s why. The results are given in the table 3. Tmin & Tmax .

6 α=0. The incident fluxes are computed though the same previously used module. 3.4 0.1 dTmin d 0.1 α=0.14 display the results for the hot and cold cases.2 α=0.8 0. Figures 3.7 Advanced Simulink Model As presented in the introduction.9 45 −15 40 −20 Tmax [°C] 35 Tmin [°C] 30 −25 25 −30 20 15 0 0.0 0.12 – Tmax & Tmin sensitivity to aluminum panel thermo-optical properties the values are exposed in table 3.7 α=0.1 Table 3.4 0.4 0. if the absorptivity of the rails increase(decrease) from 10%. Each PCB and face has a dedicated node in addition to the batteries’ node.8 -1.6 0.5 α=0. only its main features and results are presented. the maximal temperature will decrease(increase) from 0.8 Aluminium panel emissivity 1 Figure 3. dTmax d 0.1 dTmin dα 0. who is in charged of the MIAS subsystem.5 – Sensitivity coefficients to thermo optical properties It emerges that the aluminum panels properties have the more influence but that it yet remains relatively small.3 α=0.8 -0.2 0.0 -0. Beukelaers. For instance. However.8 α=0.8 -0.4 -0.8 C.9 -1. Preliminary thermal analysis 50 −10 α=0.5.8 1. V. develops a global simulator in Matlab Simulink environment.4 α=0.6 0.2 0.1 aluminum panels Rails aluminum frame Antennas’ panel -1.6 0. Within this framework. a more detailed simulink thermal model has been developed.8 Aluminium panel emissivity 1 −35 0 0. The value must be interpreted as the variation of the minimum/maximum temperature resulting from a 10% variation of the thermo-optical property.6 -0.1 1.1 dTmax dα 0.8 1. for the sake of conciseness.Chapter 3. The bottom plots show the evolution of the eclipse time while the two others Lionel Jacques 2nd Master in Aerospace Engineering 35 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .13 and 3.4 0.

Figure 3.13 – Advanced simulink thermal model hot case results Figure 3. Figure 3.13 shows the reduction of eclipse time and the resulting transient period lasting four to three orbits before the hot case is achieved. Preliminary thermal analysis represent the evolution of the temperature of the PCBs and faces.Chapter 3.14 – Advanced simulink thermal model cold case results Lionel Jacques 2nd Master in Aerospace Engineering 36 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

Moreover. will be one of the key driver of the thermal design in the hot case as well as in the cold case. On another hand. which has the narrower allowable temperature range. the available surface area for thermal control coating is low and their individual effect on the extreme reached temperatures are quite small. Preliminary thermal analysis Summary The preliminary analysis already suggests that the battery.Chapter 3. However. hot spots could occur in the hot case but are not observable here. accurate data about the optical-properties of the outer surfaces should be gathered Lionel Jacques 2nd Master in Aerospace Engineering 37 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

The global thermal model is thus divided up into two separated models with different purposes : the Geometric Mathematical Model (GMM) and the Thermal Mathematical Model (TMM). This choice is motivated by relatively low parametrization of ESATAN/ESARAD models. namely the Simplified Thermal Model (STM). using batch mode. It is based upon the lumped parameter method. 38 . The TMM includes conductive links and heat capacity and is solved within ESATAN to predict the temperatures of the spacecraft. ESARAD/ESATAN are then launched within Matlab. is to compute the radiative coupling between its different nodes and the environmental thermal fluxes along the orbit which are then input into the TMM. GMM and TMM input files are written within Matlab. All output files are also post-processed within Matlab. except for model checking. all parametrized. The conductive network links and nodal heat capacities are calculated by hand. Figure 4. 4. These two stages are performed respectively with ESARAD and ESATAN softwares. based on spacecraft and environmental data.1 shows the thermal model processing flow chart. parametric studies are easier to perform since all the process is Matlab controlled. the use of ESATAN/ESARAD softwares seemed necessary even for the development of a simple multi-nodes model. For our model creation. Then. This chapter includes a brief description of these softwares and thermal modeling. The particular modeling flow chart will be described.1 Thermal modeling with ESATAN & ESARAD The global thermal analysis is performed in two stages : the radiative analysis and the thermal analysis. GMM’s goal.Simplified Thermal Model 4 After this first preliminary analysis. based on material properties and geometrical configuration. The aim of this model is not to give accurate temperature distributions of PCBs but rather a first guess of the temperature of OUFTI-1’s main units which was not available in the previous single-node model. By this way. ESA’s standard tools for thermal radiation and thermal analysis. created within ESARAD. The fundamental assumption is that it considers isothermal nodes in a thermal network (electrical analogy). it will be used for the creation of our first multi-nodes model. pre/post-processing is done in Matlab environment.

1 – Thermal model processing ow chart Lionel Jacques 2nd Master in Aerospace Engineering 39 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . QI & C Matlab TMM ESATAN Temperatures Matlab Post processing Figure 4.Chapter 4. Simplified Thermal Model Geometry. Environment Matlab GMM ESARAD GR & Fluxes GL.

OBC2.OBC . Spacers and PCB connectors have not been modelled. EPS. for all face’s layers except solar cells. This is a relatively good assumption seeing the small thickness and high conductivity of aluminum. .EPS . COM). representing the average temperature of the PCB. One node per PCB (OBC.Chapter 4. The glue is neglected.6 4. i 6= 4 i = 1. One node for the two batteries. precise measurements are not affordable for a student project. Neglecting spacers is a good assumption seeing their low contribution in the radiative model (low area and low emissivity of aluminum) and low capacitance for the TMM. their heat capacity will be taken into account and included to the PCBs. data have been collected from many sources and compared in order to choose the most realistic values as possible. It is made up of 18 nodes : One node per face : each face is considered as one node and so isothermal. Batteries’ PCB.EPS2 . We call layer the different elements constituting the face along its thickness: successively the aluminum frame.. Nevertheless. this first multiple-nodes model remains relatively simple. Here is the numbering convention adopted : . considered just as conductive links. .ith face solar cells .2. the aluminum panel and the solar cells. Simplified Thermal Model 4. The assumption about the connectors is stronger : in the GMM.ith Face ..BAT . Therefore.2.COM .OBC2 . EPS2.2 illustrates the nodal breakdown and GMM. 4.2 Geometric Mathematical Model For a good understanding and easy interpretation of heat flow paths. One node for each pair of solar cells (two per face).1 Nodal breakdown Figure 4.2 Thermo-optical properties Concerning thermo-optical properties. 6. radiation inside the CubeSat is less influent than conduction. the view factors between face 5 and PCBs are artificially increased but as it will be shown.Batteries 10i 30i 10000 12000 14000 15000 16000 18000 20000 i = 1 . Recall that the CubeSat’s rails are hard anodized to prevent Lionel Jacques 2nd Master in Aerospace Engineering 40 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

the CubeSat with one aluminum panel before integration. solar cells have not been directly integrated on the panels but a Kapton R foil is used as insulator and completely covers the panels.3 – On the left. the MIL-spec reference of the anodized process is MIL-A-8625F. CLASS 1 while the one of alodyned process is MIL-DTL5541F Type II. and did optical properties measurements of the integrated panels. the panel integrated with the cells by EADS Astrium 4.2. Both processes will be studied in details in the next chapter. OUFTI-1 is selected for the Vega Maiden Flight and the only defined orbital parameters are : to Adam Reif from Pumpkin.3 Orbit & attitude As already said in the beginning of this work. Simplified Thermal Model Figure 4. aluminum panels’ surface treatment is also Alodine (1200). Data TM for the kapton covered panels comes directly from EADS Astrium. CLASS 1A 1 According Lionel Jacques 2nd Master in Aerospace Engineering 41 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Figure 4. For insulation reasons. On the right. who performed the solar cells integration on the aluminum panels. TYPE III.2 – Geometric Mathematical Model nodal breakdown in ESARAD galling while the rest of the frame is alodyned to enhance electrical conductivity1 .3 shows the panels before and after solar cells integration.Chapter 4. Figure 4.

a permanent magnet will be used to align the CubeSat on Earth’s magnetic field and hysteretic materials to damp the remaining rotation energy. As described in the ADCS subsystem presentation. Due to the telecommunication payload.08 0. this attitude has been simplified by using a planet centered Local Orbit Coordinate System (LOCS). described on the figure 4. The LOCS already performing one revolution along the orbit.85 0. the prescribed rotation rate is 360 deg/sec where T is T the period of the orbit.1 – Thermo Optical Properties its apogee: 354 km. For a perfect Lionel Jacques 2nd Master in Aerospace Engineering 42 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .08 0. inside aluminum panels.86 0. outside Solar cells PCBs Battery pack 5052 aluminum alloy 7075 aluminum alloy 7075 alloy aluminum Hard anodized Alodine 1200 2mil Kapton R foil on Alodine 1200 anti reflective coating - α 0.8 (a).81 0.2.1 0. [7] [64] [21] Notes : (a) : EADS Astrium measurements Table 4. In ESARAD. Simplified Thermal Model Part aluminum frame Material 5052 aluminum alloy Thermal finish Alodine aluminum frame rails aluminum panels.87 0.1 0.07 0.08 0.1 0.15 0. MCRT method is based on a statistical approach. the alignment on Earth’s magnetic field is approximated by prescribing a constant rotation speed around y axis such as the CubeSat perform two revolutions during one orbit.15 0.86 0.8 0.86 0. as the orbit inclination is relatively high.Chapter 4.15 0. its perigee: 1447 km and its inclination: 71 . 4. Within this coordinate system. also used in realistic rendering.81 reference(s) [27] [11] [57] [20] [24] [50] [49] [43] [45] [8] [43] (a) triple junction GaAs cells FR4 plastic 0. the satellite does not require a precise attitude control so that this will be a passive one.4.86 0.08 0.4 Radiative coupling The calculation of view factors and direct heat fluxes is performed in ESARAD by the Monte Carlo ray-tracing (MCRT).87 0.88 0.88 0.91 - 0.

obtained at a cost.6. perpendicular to +Z and in the direction of the velocity. the estimated value evolution goes randomly within a band whose width is inversely proportional to the square root of the number of rays fired. especially when compared with the alternative matrix methods. is randomly determined. This is especially true for models with low-emissivity should the number of multireflections is accuracy. figure 4. In athe pseudo-random convert into format which is more suitable for ray defining nutshell. external faces have no radiative coupling so that environmental fluxes calculation does not require many rays. and transmission are thus modelled exactly. are classified with respect to these volumetricto different voxels).the the model’s radiative of faces different seed values will lead cells (known as results.5Figure 7-2 Evolution of MCRT-calculated entities manual [2] – MCRT convergence from ESARAD user 7. from ESARAD Figure 4. This is the basis of the Monte Carlo from a finite method.4 – other than those derived from theCoordinate used to describe the ray emission and the ray/face interaction. ESARAD pre-processes the geometric model to fired fromiteacha face and the seed. a user input tracing. sample of rays. is absolutely or heat to predict the evolution the estimated value within the band finite random by averaging of theresults obtained from a (see Figure 7-2). However.1. the only definitive statement that can be made is that the evolution of the estimated path of anwithin a band whose width is inversely proportional to the square root of the value is infinity of rays is not possible. On another hand. each ray has its own path. True-Sun-oriented or Sun-oriented.4. Firing and following the results. Diffuse emission and reflection. This random evolution could lead to some mistakes and a convergence analysis was performed. Lionel Jacques 2nd Master in Aerospace Engineering 43 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . MCRT.2. and North X Y Z page 7-5 of rays. Accurate MCRT runs tend to be time consuming. The individual history of each ray. however. an estimate of the radiative couplings or heat fluxes can be made by averaging the Reference plane results obtained Reference line random sample of rays. therefore. A disadvantage of MCRT method is that required calculation time increase as the face emissivity/absorptivity decrease since the rays are reflected more times before being absorbed. seed1 ke = ------N theoretical value seed2 Figure 4. figure 4. Simplified Thermal Model PLANET_ORIENTED A planet-centred orbit may be planet-oriented. essentially a stochastic Y method.2. By setting the number of rays to 10000 (or more). It allows the modelling of the thermal radiative behaviour of the system without introducing any special assumptions or Figure 6-9 Planet-oriented LOCS restrictions.1 MCRT pre-processing The two control parameters of MCRT method are the number of rays When you perform an MCRT calculation.1 Planet-centred orbits Chapter 4. that is its emission point.User Manual The high fidelity of the method is. Due to and stochastic nature initial distribution of broken down into volume elements. X MCRT is a very powerfulZsimulation method. emission direction and ray/face interaction. specular reflection. +X is in the orbital plane. Concerning our GMM. MCRT being a the radiativeitcouplings impossible fluxes made number of the estimation of random process. MCRT is. sincebe fired from each face since dramatically increased. Concerning the accuracy of the reflected by any other MCRT runs is theirin the model. Planet centered Local Orbit idealisation System (LOCS). the space occupied by the model is rays’ orientation. as shown by Figure 4.6 illustrates the convergence of the radiative exchange factor (REF) between the EPS PCB and face 1 for different seeds.6 shows that the relative error is about 1% (or below). an infinite number of rays faces. +Z points towards the local zenith. The statistical feature of MCRT comes fromrays fired. its own initial direction and may reach and be Another characteristic of radiative face randomness.5.

This means that temperature gradients inside a face are assumed to be low and therefore neglected : the face is considered isothermal. The equivalent conductance Q is easily deduced from GL = ∆T . These values do not take into account the contact resistance in the links with faces 3 and 6. log scale 5 10 6 Figure 4. conductive links calculation is not immediate.5 error on REF between face 1 and EPS PCB [%] 2 1. TMM involves conductive links network and nodal heat capacities definition and internal heat load distribution. Hence.6 – Radiative exchange factor convergence with number of rays fired (MCRT) 4. The method used is the following one : to compute the equivalent coductive link between two faces. The 12 conductances are exposed in the table 4.2. These tasks are made by hand within Matlab and described for each group of nodes in the next paragraphs. perpendicular to the flow path. Contact results from the fact that faces 3 and 6 Lionel Jacques 2nd Master in Aerospace Engineering 44 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . This is an approximation : the contribution of aluminum panels to the conductance between faces has been neglected.3 Thermal Mathematical Model Now that the GMM is defined and ready for computation. The conductive flow path between the faces only goes through the aluminum frame since there is no other contact. each of them is cut in two in the middle. the second stage of the thermal analysis consist in the TMM creation. each face of the CubeSat is modelled by one node. Seeing the relatively complex geometry of the frame. 4. Figure 4. A temperature gradient is imposed across the two half faces and SamcefField is then able to compute the resulting flux proportional to the temperature gradient.5 0 3 10 10 4 10 Number of rays fired.1 The structure As previously described. Simplified Thermal Model 3 seed 1 seed 2 seed 3 2.7 shows the temperature distribution across the face 1 and 3 for a gradient of 10 C. for each of the 12 links. a finite element model has been created within SamcefField.5 1 0.3.Chapter 4.

988 10 2 5.465 10 2 6.2830 c [J/kg. Properties of these alloys are given in the table 4.465 10 2 5. These phenomena will be studied and taken into account in a more detailed model : the low level model.2 – Equivalent conductance between aluminum frame faces are fixed on the lateral frame with screws : one for each face.1002 913 .K] 140 .789 10 2 5.3 – aluminum alloys properties Using these properties. The frame is made of 5052 aluminum alloy while 7075 aluminum alloy was used for aluminum panels.924 10 2 8.K] 963 . Alloy 5052 7075 ρ [kg/m3 ] 2672 .137 Table 4.465 10 2 5.496 10 2 5.465 10 2 Table 4. Lionel Jacques 2nd Master in Aerospace Engineering 45 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . the nodal capacitances of the faces are computed by adding the capacitances of the frame face and associated panel.2698 2770 .979 k [W/m.519 10 2 faces GL26 GL34 GL35 GL45 GL46 GL56 GL [W/K] 5.Chapter 4.465 10 2 5.496 10 2 5.7 – Evaluation of the conductive link of aluminum frame through a finite element analysis within SamcefField Faces GL12 GL13 GL15 GL16 GL23 GL24 GL [W/K] 5. The two assumptions tends to balance each other since that neglecting contact overestimate the conductance while neglecting panels conduction contribution underestimate the total conductance. except for face 4 fixed with 3 screws because of the ports. Simplified Thermal Model Figure 4.465 10 2 5.152 131 .3 and comes from the CES EduPack software.

4 – Nodal capacitances 4. In a first design. meaning that it refers to the incident power and not the absorbed one.5 46. The equilibrium gives: Qin = (1 α) Qin + Q Ther + Q Elec Lionel Jacques 2nd Master in Aerospace Engineering 46 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .5 58. Simplified Thermal Model Node 301 302 303 304 305 306 C [J/K] 46. The energy collected by the cells is not directly available in ESARAD but can be easily obtained from the power balance of a solar cell (figure 4.8). equal to the projected area A multiplied by the solar constant Cs and η is the solar cell efficiency.5 63. The efficiency available in the data sheets is indeed the AM0 efficiency.1 46. Ledent and P. a dissipation circuit has been designed by P.8 – Solar cell power balance Qin is the incident power. the extra power is thus dissipated through electrical resistances on the EPS PCB.Chapter 4.3.9 48. The only way of dissipation in space is heat.2 The solar cells Electrical power management As the SwissCube. As the CubeSat’s attitude is passively controlled. Thirion in order to deal with the extra power produced by solar cells when the batteries are charged and the required power is lower than collected one. no face is pointing to deep space and the use of a radiator to ensure heat rejection is not to be foreseen. d    Qin = ACs   Qre f l = (1 α) Qin d   d   d   d ‚ d   d d     d   Q Ther d Q Elec = η ( T ) Qin   d ©   ‚ d Figure 4.9 Table 4.

3 The PCBs stack As explained previously. Lionel Jacques 2nd Master in Aerospace Engineering 47 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . the equivalent thermal resistance between the cell and the face becomes : 1 GLcell. a specific heat of about 700 [J/KgK] [59] and weights 86 [mg/cm2 ][7]. The heat capacity of the PCBs is evaluate by multiplying their mass (about 70g. The equivalent specific heat is determined with the help of the online calculator developed by Frigus Primore [46]. including PCB and connector) with a equivalent specific heat. Simplified Thermal Model Therefore.39 TM Kapton RTV S 691 R Table 4. the heat generated by the cells is given by: Q Ther = (α η ( T )) Qin But ESARAD gives us the absorbed power : Q ESARAD = αQin and here is the electrical collected power Q Elec expressed in function of Q ESARAD : η (T ) Q ESARAD α In ESATAN.5 – Thermal conductivity of Kapton [14] and RTV S691 adhesive [18] The cells are 150 microns thick [7]. The layout of the solar cell integration made by EADS Astrium is the following one : aluminum panels have been covered with 50 microns (2mil) thick Kapton R foil for insulation reason. inside the CubeSat. f ace = Kapton R and RTV S 691 properties are given in the table 4. 4. f ace 1 Acell tKapton tcell t t t + adhesive + + adhesive + aluminum k cell k adhesive k Kapton k adhesive k aluminum = Rcell.12 0. partially or not. each PCB is modelled with only one node.Chapter 4. The laydown adhesive is RTV S 691 (silicone adhesive) for the Kapton R foil and for the solar cells and is 80 microns thick.3. where dissipation occurs. representing the mean temperature of the PCB and this model will not be able to reveal local hot spots inside the PCBs.5 k [W/m. have a thermal conductivity of more or less 100 [W/mK] [59]. From this layout. this power is subtracted at the cells and reinjected. Q Elec = η ( T ) Qin = Solar cells' model Solar cells have a dedicated node with its own capacitance and its link with the corresponding face.K] 0.

169 reference ECSS Q-70-71A[18] CES EduPack Phosphor bronze 6061-T6 aluminum Table 4. Figure 4. perfect contact is assumed. Thermal conductivities of phosphor bronze and 6061-T6 aluminum are given in table 4.5mm (exterior). Simplified Thermal Model Inter PCBs links Concerning the conductive links between the PCBs.6 : k [W/m.2 104 A1pins k pins 104 0.35 10 6 160. there are two main heat flow paths : through the spacers and through the connector. According to the PC/104 specification [41]. Again. The conductive link through is not easy to evaluate and. the connector is made up of 104 phosphor bronze pins and only pins contributes to the link since connector’s housing height is such as there is no contact with the above PCB. OBC PCB is supported by four stainless steel threaded loose fasteners (in which the endless screws are screwed) inserted in the bottom frame face.3mm (interior) and 4. Once again.11).9 illustrates the temperature distribution inside a midplane standoff when a gradient of 10 C is imposed between the two perpendicular faces involved. perfect contact is assumed between the PBC and the frame.3.K] 75 152 . a finite element model has been created within SamcefField. The Lionel Jacques 2nd Master in Aerospace Engineering 48 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . The top links are ensured bu the "midplane standoffs" as presented on the exploded view in chapter 2.Chapter 4. PCBs stack links with the structure The PCBs stack has 8 contact point with the external structure : four at the top with the COM PCB and 4 at the bottom with OBC PCB.5 = = 1. The spacers are made of 6061T6 aluminum and the contact between spacers and PCBs is considered to be perfect. Concerning the bottom link of the PCBs stack. as for the aluminum frame links. 4. The equivalent thermal conductance is obtained by dividing the resulting flux by the difference of temperatures.508 10 6 75 This means that conduction through the spacers is as well important as conduction through the connectors. Two midplanes ensure the fixation with the face 5 and two with face 2. Conduction through the M3 endless screws is neglected seeing the clearance with the PCB holes and spacers.4 The batteries The two LiPo batteries are located on a secondary smaller PCB itself fixed on the EPS PCB as shown by the exploded view (figure 1. A spacer is a small hollow aluminum cylinder with diameters of 3.6 – PCBs links materials thermal conductivities The ratio between the link through spacers and the link through connector is: 4 Sspacers k spacers 4 7.

58 1.OBC2 OBC2 . The material of these spacers is not yet defined and is kept as parameter.42 1.face 2 COM .Chapter 4.58 0.COM COM . the internal power is distributed between the two PCBs containing the most dissipative components : COM (amplifier) and EPS (dissipation circuit.face 5 GL [W/K] 0. we used the one of Varta PoLiFlex R available in the SwissCube Phase C technical report [35].EPS2 EPS2 .42 Table 4.12 0.Hot case : maximum environmental fluxes. Simplified Thermal Model Figure 4.35 0.BAT PCB EPS . Concerning the specific heat of the batteries. thermal analysis is subjected to a worst cases approach.4 Worst cases definition As introduced in the previous chapter. aluminum conductivity will be assumed as initial value.58 0. The assumptions about cold and hot cases are the same as the one described in the preliminary analysis : . voltage converter).OBC OBC . 4. Now that PCB are modelled. keeping in mind that the model used is not yet defined. orbit permanently illuminated.35 0.EPS EPS . Without Lionel Jacques 2nd Master in Aerospace Engineering 49 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .9 – Conductive link computation through a finite element analysis within SamcefField of a midplane standoff Structure . electrical collected power entirely dissipated into heat.7 – PCBs links conductive link between the batteries’ PCB and the EPS PCB is also ensured by four spacers.

Concerning the solar cells efficiency. is already high (0. As the absorptivity of Kapton R . The effect of thermo-optical properties degradation is only taken into account for the Kapton R used on the panels. this simulation confirms the result obtained in the preliminary analysis : the safety margin is narrow and other solutions may be foreseen.5 Results Figure 4. considering the maximal allowable temperature of the batteries (45 C). its absorptivity increasing over time. But. since aluminum surface treatments are relatively stable over a lifetime such as the one expected for OUFTI-1 (about one year). here set at one degree per second. decreasing over time. It shows that the EPS2 PCB is quite cooler than the EPS PCB since no power is dissipated on EPS2. the dissipated energy on EPS2 will be smaller in any case. a diminution of 10% is considered : 30% BOL and 27% EOL. 4. it remains cooler than the others mainly because it is linked with two opposite faces (2 and 5) and when the first one is in sunlight. This solution will be studied in details in the low level model. the other is consequently turned to deep space. The other reasons for which EPS2 is cooler is deduced from the same graph. The total power is represented by the black dotted line and the effective thermal power.10 shows the different contributions to the total absorbed power during one orbit.87) the End-OfLife value is set to 0. It shows that even if half of the collected power is dissipated on the COM PCB. nearly ten times more con- 4. obtained by subtracting the electrical power collected by the cells from the total power. according to [33] and [55]. A possibility could then to attach the batteries’ PCB to EPS2 PCB instead of EPS PCB.Chapter 4. an equal distribution between COM and EPS has been assumed. The other fact involving cooler temperatures of COM and EPS2 PCBs is that the conductive links with the structure is greater than the one linking the OBC to the structure (the midplane standoff are made of aluminum. A possible solution could be imagined by analyzing the graph in the bottom left corner.11 shows associated ESATAN results.Cold case : minimum environmental fluxes.1 Hot case Lionel Jacques 2nd Master in Aerospace Engineering 50 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . They are all comprise between 30 and 35 C. all the electrical power is redistributed inside the CubeSat as internal heat loads so that the total power (dotted line) finally contributes to the heating of the CubeSat. measured by EADS Astrium. It is obvious that they are quite different from those exposed in the preliminary analysis. . The rapid variations are the consequence of rotation of the CubeSat around its axis aligned on Earth’s magnetic field. Actually. and the solar cells efficiency. maximum eclipse time (apogee in eclipse involving more than 35 min in eclipse for a period of 104 min). As explained above.9. is the continuous black line.5. no internal dissipation. Simplified Thermal Model accurate information about these components. The bar plot on the right shows the maximal temperature reached at the main nodes during the orbit. Figure 4.

2°C 29.59°C 34.82°C 36.76°C 36.74°C 36.11 – Hot case temperatures Lionel Jacques 2nd Master in Aerospace Engineering 51 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .78°C 34.32°C 34.Chapter 4.63°C 35.10 – Hot case absorbed power 40 Batteries 36.98°C 33.15°C 34.17°C 34.55°C 33.18°C 30. Simplified Thermal Model 20 Solar Albedo Earth IR Total Total − electrical Electrical Absorbed power [w] 15 10 5 0 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 4.12°C Temperature [°C] 30 COM PCB EPS2 PCB 20 Face 1 Face 2 Face 3 Face 4 Face 5 Face 6 EPS PCB OBC2 PCB OBC PCB Face 6 Face 5 Face 4 Face 3 10 0 0 20 40 60 Time [min] 80 100 40 Face 2 Face 1 Solar Cells 6 Temperature [°C] 35 30 OBC OBC2 EPS EPS2 COM Batteries Solar Cells 5 Solar Cells 3 Solar Cells 2 Solar Cells 1 25 20 0 20 40 60 Time [min] 80 100 0 10 20 30 40 Maximal Temperature [°C] 50 Figure 4.64°C 36.93°C 36.52°C 30.

5 and 6 while faces 1 and 4 spends more time in sunlight or shadow. The rotation about this axis is clearly visible through the fast variations of faces 2. EPS2) are less sensitive to external temperature variations than OBC and COM PCB. one can see that internal components undergo relatively lower temperature variations than the external structure.Chapter 4. Globally. This is intuitive since central PCBs are more insulated from the external structure. Face 1 and 4 have a different comportment than the others : this is a consequence of the passive alignment of one axis of the CubeSat on Earth’s magnetic field. Moreover. A steady state analysis considering the mean incident fluxes would then be suitable and more computation-time effective. Simplified Thermal Model ductive that stainless steel used for bottom spacers.3.7). depending on the initial condition after being released from the P-POD and on the moment generated by the deployment of the antennas. Figure 4. the rotation rate considered here is 1 deg/s. ADCS simulations [25] shows that the CubeSat will certainly undergo higher rotation rates.12 illustrates the consequence of a rotation rate of 5 deg/s. The variations are then too fast compared to the thermal inertia of the faces and PCBs. especially for the next model. One must keep in mind that those temperatures are averaged temperatures which are then not able to represent of local hot spot that may occur. Another intuitive consequence of the PCB configuration is that central PCBs (EPS.12 – Evolution of the temperatures during one orbit for a rotation rate of 5 deg/s Lionel Jacques 2nd Master in Aerospace Engineering 52 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . especially on the PCBs. as it was indicated in the table 4. much more complex. 40 35 Temperature [°C] 30 25 20 15 10 5 0 0 10 20 30 40 50 Time [min] 60 70 80 90 Face 1 Face 2 Face 3 Face 4 Face 5 Face 6 100 40 Temperature [°C] 35 30 OBC OBC2 EPS EPS2 COM Batteries 25 20 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 4. The graph in the top right corner illustrates the temperature evolution of the faces. These ones do not undergo anymore fast variations and are almost constant.

14. excluding the evident eclipse time. it is time to examine the cold one.2 Cold case Now that the hot case has been analyzed. The off-peaks visible in the absorbed power are turned here into peaks: shortly after 10 minutes. The second off-peak is lower than the first one because of the lower absorptivity of face 4. the projected area being much lower. But the important fact is not there: the temperature of the batteries drops below -20 C while its allowable minimal temperature in charge is 0 C! Obviously. 25 Solar Albedo Earth IR Total Total − electrical Electrical 20 Absorbed power [w] 15 10 5 0 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 4. Simplified Thermal Model 4. a heater will be needed to raise the temperature of the batteries. no transverse conduction is taken into account in the link between the EPS PCB and the PCB sustaining the batteries. Once again. the evolution of absorbed powers along the orbit is represented at figure 4.Chapter 4.13 – Cold case absorbed power Concerning the temperatures. the assumed material of the spacers between the EPS and 1 As it has been studied in the preliminary analysis.5. the upper left graph shows that face 1 undergoes a rise of its temperature. Furthermore. But one must pay attention to the limits of the model : as each PCB is modelled with one node. not covered with solar cells. is a higher albedo since the CubeSat get over the subsolar point1 . The second feature visible is the presence of two off-peaks around 10 and 90 minutes. results are illustrated by the figure 4. These off-peaks occurs when the face 1 (first one) or 4 (second one) is perpendicular to sun rays. A first difference with the hot case. the higher albedo present in the cold case and not in the hot case does not affect significantly our worst case definition Lionel Jacques 2nd Master in Aerospace Engineering 53 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .13.

03°C −28.05°C −28.47°C −21. .57°C −30.. L increasing the length or reducing the area would have the same effect but structural constraints must be taken into account (resistance. These two assumptions are not favorable to the batteries.44°C −30. ).7°C −32.62°C −24.12°C −40 −30 −20 −10 Minimal Temperature [°C] 0 5 0 Temperature [°C] −5 −10 −15 −20 −25 −30 −35 0 20 40 60 Time [min] 80 100 OBC OBC2 EPS EPS2 COM Batteries Face 2 Face 1 Solar Cells 6 Solar Cells 5 Solar Cells 3 Solar Cells 2 Solar Cells 1 −50 Figure 4.09°C −28.Chapter 4. But this was for EPS-BAT spacers in aluminum. Using CES EduPack.62°C −28.4°C −27.15°C −32. rigidity.5. materials having very low thermal conductivity are thermoplastics or thermosets but not compatible with structural aspects. The scope of this sensitivity analysis is then to evaluate the effect of reducing the thermal conductivity of these spacers. Lionel Jacques 2nd Master in Aerospace Engineering 54 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .5°C −28. Again. density. 4.8°C −22. The graphic on the right shows that the efficiency of this effect is maximum between 1 and 100 W/mK.14 – Cold case temperatures BAT PCB is aluminum which is a good conductor.52°C −30. titanium and titanium alloys appeared to be a good choice and having a thermal conductivity about 10 W/mK with a high Young’s modulus and high tensile strength.88°C −21. Simplified Thermal Model 20 10 Temperature [°C] 0 −10 −20 −30 −40 0 20 40 60 Time [min] 80 100 Face 1 Face 2 Face 3 Face 4 Face 5 Face 6 Batteries COM PCB EPS2 PCB EPS PCB OBC2 PCB OBC PCB Face 6 Face 5 Face 4 Face 3 −21. the COM PCB is cooler than the others for the same reasons that those previously presented.14°C −28.3 Sensitivity analysis Cold case has shown that the batteries had approximatively the same temperature as the EPS PCB one.15 shows that insulating the batteries has a positive effect: reducing the conductivity from 150 to 5 W/mK increase the minimal temperature reached by the batteries from -21 to -17 C. The influence of the EPS-BAT spacers conductivity Again. Figure 4. . Considering the definition of a conductive link GL = kS .

several assumptions has to be removed in order to properly design suitable solutions. Lionel Jacques 2nd Master in Aerospace Engineering 55 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . before presenting the Detailed Thermal Model.1 [W/mK] k = 1 [W/mK] k = 10 [W/mK] k = 50 [W/mK] k = 100 [W/mK] k = 400 [W/mK] Batteries minimal temperature [°C] −16 −17 −5 Batteries temperature [°C] −18 −10 −19 −15 −20 −20 −21 −25 0 20 40 60 Time [min] 80 100 −22 10 10 EPS−BAT spacers’ conductivity [W/mK]. and active with small heaters. a combined solution. Simplified Thermal Model 0 k = 0. some measurements are required and discussed in the next chapter. through batteries insulation. However. This is the aim of the Detailed Thermal Model. log scale 0 2 Figure 4.Chapter 4. further analysis will be developed with the low level model because some important effects neglected here such as transverse conduction through PCBs will be taken into account. will certainly be required to maintain the batteries in their safe range of temperatures. Summary This Simplified Thermal Model has allowed to evaluate the average temperatures of each PCB and confirmed the potential problems considering the batteries for both hot and cold cases. both passive. Nevertheless. even if it allowed to foresee some possible solutions to these problems. However.15 – In uence of the EPS-BAT spacers conductivity on batteries’ temperature Once again.

several experiments will be done to determine different properties of the aluminum frame and the batteries. 5. It acts like an hysteresis and prevent the heaters from too fast pulsing. Indeed. The thermal design of a satellite involves a good knowledge of thermo-optical and materials properties. the data used in the previous models were based on SwissCube model but their batteries are packed in small aluminum boxes. only few measurements can be performed. 5. the combination of thermal capacitance and conductivity gives us a delay between the time the heater is turned on and the sensor response. Moreover. notably about the batteries. Panasonic) are all of approximately the same size and made upon the same technology. as already spotted. But in a student project. Yet. The second measurement focuses on the thermal emissivity of the aluminum frame’s two coatings : alodine and hard anodizing.1 Why and which measurements ? Li-Po batteries thermal properties are not well known and difficult to find.2.Measurements 5 This chapter describes several measurements performed at the Centre Spatial de Liège. they are the more critical unit seeing their narrow range of temperatures and measurements seemed necessary.2 The Battery The goal of this test is thus to evaluate the thermal capacitance and transverse conductivity. In this chapter. 56 . Kokam. These two properties are crucial when designing heaters : how much power will be required to maintain the two batteries above a given threshold and where to put the control sensor.1 Experimental setup This test will be performed on a battery Kokam SLPB-554374-H which is representative since the foreseen models (Varta. around the screws. That is the reason why the first experiment concerns the determination of the thermal properties (conductivity and the specific heat) of a typical LiPo battery such as the one probably used in our CubeSat. This test is based on a thermographic measurement. This is also the case for the third one about the thermal contact between the top and bottom faces with the lateral frame. 5. as it has been shown previously.

a Simulink model TM fitting at best the setup has been created. heat and temperature sources and sensors etc. As all the test is performed under ambient conditions. one for the interface plate and one for the bench. To reduce errors. Unfortunately. Seeing its large capacitance. Acquisition is made through thermocouples and PT100 sensors linked to a PC equipped with a Keithley 2700 datalogger. Thermal compound1 is used between the interface plate and the battery to ensure a good contact. As shown on the figure 5. 5. This model takes also into 1 Wakefield TM Engineering Thermal Compound n 126-4 2 Multi Layer Insulation with Dacron netting or "bridal veil" between the layers to reduce contact Lionel Jacques 2nd Master in Aerospace Engineering 57 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Figure 5.1 – Battery test setup Concerning the heater used is a Minco HK5164R39. the battery is then strongly pressed on the interface plate with two pieces of CFRP. . the heater does not cover the entire area of the battery as shown on the right picture of the figure 5. the temperature of the battery bottom side is no more imposed to model the observed local heating of the interface plate and bench.Chapter 5. It contains four thermal masses : two for the top and bottom side of the battery.2. multiple consecutive heating cycles are performed and the model will be adjusted for all the cycles. seeing the results. By this way. . conductive/radiative/convective links. heaters are covered with aluminum tape to prevent heat flux from escaping radiatively and all the setup is covered with MLI2 again to diminish radiative losses and reduced convective exchanges. CFRP has been chosen for its high rigidity to thermal conductivity ratio. the temperature of the heated side will rise up to its constant steady state value.2. Our model is shown on the figure 5. the temperature of the bench is assumed to remain constant and equal to the ambient temperature.1. Indeed. the battery is fixed on an interface plate which is itself strongly fixed with bolts on a massive steel bench.2 The model In order to determine the two desired properties.1 but this will be taken into account in the model. allowing to create transient thermal model with simple constitutive elements: thermal masses. Simscape toolbox available TM TM in Matlab Simulink R2008 is used.2. To impose the temperature of the "rear side". Measurements The principle of the test is to heat one side of the battery while imposing the temperature of the opposite side.

Chapter 5. one decided to consider that the heat flux propagates through a reduced area equal to mean area between the heater and the bottom side ( 65+50 40+25 [mm]). 2 on the bottom side and 2 on the interface and bench Lionel Jacques 2nd Master in Aerospace Engineering 58 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . t fcn Power P S PS Clock S A B A B T PS S B Battery Top 1 Out1 B T f(x)=0 A A PS S B Battery Bottom A A B T PS S 2 Out2 A B Interface Plate Bench Figure 5. The measured temperatures presented at the figure 5. if the bench had not been modelled and replaced by a fixed temperature. By adjusting the different parameters of the model. Measurements account the conduction through the thermal compound between the bottom side and the interface plate. At the same time. Intuitively.3 results from an averaging between the temperatures measured by the different sensors.2. The same consecutive powers really injected in the heater is set as input and the three measured temperatures are set in output: battery both sides and the interface plate. the global heating of the system could not have been represented (the temperature ad the end of the test is greater than the initial one because the bench absorbed some energy and did not have the time to release it seeing its large thermal inertia). As the heater does not covered all the area and thanks to the different sensors1 used. The thermal inertia of the interface plate has its importance: it acquires and releases energy and contributes to the damping of the system. it has been observed that the temperature of the top side was of course not uniform.2 – Battery test Simscape model TM 5.5 5. the measured temperatures are finally 1 We had 4 sensors on the top side. It is such as the heat 2 2 flux propagates according to a truncated pyramid shape whose bottom face would be the bottom side and top face the heater.6 [mm]) which leads to a volume of 13000 [mm3 ] for a weight of 34g.3 Model adjustment & results The SLPB-554374 real dimensions differed somehow from the one mentioned in the datasheet : the real ones were 65 40 5 [mm] (instead of 70 42.

55 [W/mK] (through the reduced area which is equivalent to 1. anode and packaging shell have lower specific heat. The frame has two different surface treatments: the rail are hard anodized and the remaining surface is alodyned and here are a few words about these two processes.11 [W/mK] through the entire area). This leads us to believe that the determined value is not too far from reality.Chapter 5. Although the internal composition of the battery is difficult to known. anodized surfaces have a enhanced protective oxLionel Jacques 2nd Master in Aerospace Engineering 59 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . measured Bottom face. it appears that the electrolyte of Li-Po batteries is held in polymer composites such as polyethyleneoxide (PEO) or polyacrylonitrile.3 – Battery test comparison between measurements and adjusted model This involves a heat capacity of 45. 60 55 Top face. measured Interface plate. The purpose of both coatings is to improve corrosion resistance but while Alodine maintain electrical conductivity. Measurements fitted and the desired properties are determined: the specific heat is about 1350 [J/kgK] and the conductivity 1.9 [J/K]. measured The same. Such materials have high specific heat about 1800 [J/kgK].3.1 Anodizing vs Alodine The second test mentioned concerned the measurement of the emissivity of the Pumpkin structure. simulated 50 Temperature [°C] 45 40 35 30 25 20 0 10 20 30 40 Time [min] 50 60 70 Figure 5. The results were also compared with data from specific measurements on other Li-Po batteries [60]. the cathode. The representativeness of the test compared with the real configuration is also limited since larger heaters will probably be used to cover all the available area and batteries will not be fixed on aluminum plate but on CFRP (PCB).3 aluminum frame emissivity 5. One must keep in mind that the employed method relies on some assumptions : radiation and convection losses were neglected even though limiting their effects. 5. On the other hand.

4 excerpt from a NASA report about thermal control coatings. emission from the object we are looking at. Nevertheless. Alodine is a chemical conversion coating (also known as Iridite or Chromate Conversion). According to Adam Reif from Pumpkin.2 Measuring emissivity with a thermographic camera Here is a brief theoretical recall about thermographic measurements. 5.3. the rail were anodized to prevent galling (as a reminder. figure 5.5. For example. TYPE III.Chapter 5. Figure 5. Wtot . Measurements ide layer which is also a poor electrical conductor. shows how the chemical reaction time of a chemical conversion coating on aluminum affects its optical properties. CLASS 1A for alodine means that no hexavalent chromium (environment friendly) is used in the process (Type II) and that maximum protection against corrosion is provided (class 1A). the total input power collected by the camera. one must be very cautious on how to interpret the measured data. Anodizing is an electrolytic passivation process increasing the oxide layer thickness. Many thermal emissivity measurements method already exist but they often required specific setups and expensive apparatus. Lionel Jacques 2nd Master in Aerospace Engineering 60 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 .4 – Evolution of the emissivity and absorptivity of aluminum sample subjected to a conversion coating surface treatment in function of the reaction time [61] So. emissivity measurements were planned. Both treatments improve adhesion for paint primers and glues. the rails are the only part of the CubeSat in contact with he P-POD). In our case. Indeed. CLASS 1A for Alodine [39]. CLASS 1 [38] for the anodized process and MIL-DTL-5541F Type II. A simple and relatively convenient way of doing this is to use a thermographic camera. has different contributions: 1. precise data about alodine and hard anodized emissivity were quite difficult to find and optical properties are highly surface treatment dependent. here are the military specifications (MIL-spec) reference of these two processes: MIL-A-8625F. Type III CLASS 1 means non-dyed hard anodizing and Type II. As depicted in Figure 5.

3: Atmosphere. however. 2006 Figure15. 2: Object. in most cases they are fortunately small enough to be neglected. the measurement configuration is likely to be such that the risk for disturbance is obvious. by changing the viewing direction. 1: Surroundings. the contribution coming from the reflected temperature of the surroundings increases according to 1 and so does the sensitivity to reflected environment parameters. We can then write (Equation 1): the object emissivity 125 the atmosphere transmittance τ depending on the distance between the object and the camera. a196 – ENGLISH (EN) – December 21. 2: Object. from a blackbody source of temperature Assume that the received radiation power W4: Camera Tsource on short distance generates a camera output signal Usource that is proportional to the power input (power linear camera). we can use the figure below to derive a formula for the calculation of the object temperature from the calibrated camera output. The influence of these University of Liège Applied Sciences Faculty Academic Year 2008 2009 61 . 2. the atmosphere relative humidity and its temperature surroundings average reflected temperature surroundings effective emissivity Concerning. 3. Such disturbances are difficult to quantify. 1 558 071 Rev.This description of the measurement situation. In case they are not negligible. The total collected radiation power seen by the camera is then : Wtot = τWobj + (1 )τWre f l + (1 τ )Watm and the radiation power effectively emitted by the object is given by: 1 1 1 τ Wtot Wre f l Watm τ τ These equations show that to determine the effective object emitted power. Measurements Chapter 5. What has been neglected could for instance be sun light scattering in the atmosphere or stray radiation from intense radiation sources outside the field of view. the surrounding effective emissivity. is so far a fairly true description of the real conditions. A way to reduce this sensitivity is to increase the ratio between object and surroundings emitted radiation. reflected emission from ambient sources.a1 Accepting the description above. No. the camera has to subtract the two unwanted contributions. at least to a trained operator.5 – A schematic representation of the general thermographic measurement ings. These contributions involve four main parameters the user has to sup15 ply: Publ. It is then his responsibility to modify the measurement situation to avoid the disturbance e. atmosphere emission.1 A schematic representation of the general thermographic measurement situation. shielding off intense radiation sources etc. But as the emissivity decrease. 3: Atmosphere. 4: Camera situation from the ThermaCAM User Manual [21].1: SurroundFigure 5. 10400503. as illustrated in the figure below. all radiation emitted from the object is assumed to never come back so that the environment can be considered as a blackbody and its effective emissivity is one.g. The emitted radiation is proportional to the temperature at the Wobj = fourth power and the ratio is thus equal to Lionel Jacques 2nd Master in Aerospace Engineering 1 4 Tre f l 4 Tobj .

3.6 – In uence of the object temperature and emissivity for a thermographic measurement For this reason.1. noticing the resulting temperature given.8 shows the infrared picture of the structure heated up to 90 C. the more important is the influence of temperature.6 ε=0.4 ε=0.9 8 7 6 4 1− Tref l 4 Tobj 5 4 3 2 1 0 20 40 60 80 100 120 140 Object temperature Tobj [°C] 160 180 200 Figure 5. In practice.1 ε=0.3 ε=0.7. The structure was heated through a plate on which we fixed heaters. the object emissivity parameter in the software shall be varied. For example. Measurements two parameters are represented in Figure 5. Lionel Jacques 2nd Master in Aerospace Engineering 62 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . we enclose the structure with a high emissivity box of known temperature.7 ε=0. several measurements shall be performed at different temperatures.Chapter 5. for each temperature. The lower the emissivity. In order to have a uniform surrounding environment. The two considered areas for the emissivity calculation are the one enclosed by green rectangles.8. with an emissivity of 0. as shown the the right picture of the figure 5.6.3 Experimental setup The camera used was a ThermaCAM R S40 from FLIR Systems equipped with a 320 240 microbolometer array. 9 ε=0.7. If the temperature calculated by the software is equal to the one measured by the sensor then the corresponding emissivity must be close to the real one. 5. left).5 ε=0.7 and 5. We placed the camera close to the structure to limit atmosphere effects (Relative humidity was 47% and distance 30cm) and tilted the frame to avoid camera’s own emitted radiations reflection like looking in a mirror (figure 5. heating the object up to 80 C reduces the ratio from 9 to 4. close to which sensors are set as shown in figures 5.8 ε=0.2 ε=0. Figure 5.

as previously said. the emissivity leading to a zero error seems well to converge. [21]) and white paper sheet ( = 0. [21]). The left plot represents the hard anodized results while the right plot displays the Alodine ones.8.14 and 0.Chapter 5. two consecutive curves were well spaced.3. Seeing these graphs. The Alodine curves are more emissivity sensitive and that confirms the theoretical curves of Figure 5.6 showing that when the emissivity was low. the method has been validated by measuring samples whose emissivity were well known : 3 mil kapton foil on aluminum backing ( = 0. Figure 5. common aluminum plate ( = 0. measurements were performed at different temperatures and for each of them.81.9.69.1. Lionel Jacques 2nd Master in Aerospace Engineering 63 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .4 Results First.9 also confirm that as the measurement temperature increase. the emissivity is varied and a relative error can be deduced between the measured value and the infrared calculated one: TIR Tmes Tmes The evolution of this error in function of the measured temperature and emissivity is represented in Figure 5.7 – aluminum frame emissivity test setup Figure 5. Measurements Figure 5. we can finally conclude that the emissivity of the Alodine and hard anodized coatings are respectively about 0.8 – Infrared capture of the structure heated up to 90 C 5. [52]). Then.

66 0. Then. face 6 and 1.15 0.68 0.9°C 0 −5 0.2 0.9 – aluminum frame emissivity test results 5. infrared imaging is again used.7 Hard Anodized rail emissivity 0.12 0.12 0. fastened with 3 screws and on the right.75 −5 0.65 0.14 0.3°C T = 66. Measurements Error between sensor T° and IR measured T° [%] 5 T = 42.7°C T = 88.11 40 60 80 100 Temperature [°C] 120 Figure 5. Lionel Jacques 2nd Master in Aerospace Engineering 64 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .67 0. fastened with one screw The experiment is done under ambient conditions so that convective losses are absolutely not negligible and difficult to take into account.10 – Aluminum frame contacts in front of bright background.4°C T = 58.64 40 60 80 Temperature [°C] 100 Alodine emissivity 0.65 0. For this purpose. The contact phenomenon between frame elements is clearly highlighted in Figure 5.2°C T = 102.1 0.18 Alodine emissivity 0.69 0.5°C T = 52. face 6 and 4.16 0.10 which presents some screwedjoints in front of a bright background: only a fraction of the surface area contributes to heat flow path between the faces.4 aluminum frame contact resistance The third experiment concerns contact between top and bottom face screwed on the lateral frame. Figure 5.7 Hard Anodized rail emissivity 0.14 0. the aim of this experiment is not to compute accurate thermal conductances values but rather to evaluate the importance of the phenomenon.1°C T = 99°C 0 Error between sensor T° and IR measured T° [%] 5 T = 46. On the left.13 0.3°C T = 58.Chapter 5.8°C T = 115.

Figure 5.11 shows the IR pictures of the contact between faces 6 and 1 (left) and between faces 6 and 2(right). To do this.Chapter 5. The first one confirms that the contact is well localized: the area below the screw is hotter meaning that only this part of the surface contributes to the contact resistance and that constriction phenomenon occurs (shrinkage and concentration of the heat flow lines around the contact area). IR imaging can then show if the contact area is strongly localized or well spread across all the area of the bracket. the right picture is less convincing: heat flows seems to be more distributed on all the surface. On the contrary.11 – Infrared images Battery test setup In the same way. all the heat (except convective and radiative losses) must flows through the contact resistance. face 6 is heated up while face 3 is strongly fastened to a large bench (high thermal inertia) to prevent its temperature from rising. Measurements The method consists in imposing a heat flux between faces 3 and 6 through the contact. By this way. Figure 5. Lionel Jacques 2nd Master in Aerospace Engineering 65 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . the other contact areas were also studied leading to the same conclusions: contact is present but difficult to estimate and varies from one screwed-joint to another.

it demonstrated the presence of contact phenomenon between the lateral frame and bottom/top faces. This phenomenon will then be taken into account through data found in the literature and finite elements analysis. Lionel Jacques 2nd Master in Aerospace Engineering 66 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . the emissivity of the frame rails is in reality lower than the assumed values : 0. Measurements Summary The preliminary analysis and simplified thermal model showed that thermo-optical properties were important and that the batteries thermal inertia contributes largely to the global thermal of the CubeSat. For instance. Measurements allowed us to determine relatively accurate values for the emissivity of the aluminum frame and for the thermal properties of the battery.Chapter 5. These new values will be used in the newt chapter describing the detailed thermal model of OUFTI-1.8 and the heat capacity of the battery is quite greater: 1350 [J/kgK] instead of 960 [J/kgK]. Even if the third experiment did not lead to concrete values.7 instead of 0.

Actually. This means that the heat load will probably be strongly localized. analyzing and optimizing the model. All these issues suggest to develop a more precise model allowing concrete decisions and solutions.1. especially electronic circuits. But as represented in Figure 6. Both revealed the need of an active thermal control to keep the temperature of the batteries above its lower limit. their thermal inertia and dealing with the internal dissipated power distribution. But the Simplified Thermal Model (STM) was too limited to allow a convenient and realistic design of a heater. the dissipation system introduced in the previous chapter also evolved. Moreover. the STM was also unable to represent the potential hot spots due to low efficiency electronic components such as the amplifier.1 Geometric Mathematical Model Like the Simplified Thermal model and any other thermal model. 6. The new design is composed of a dissipation transistor in series with a resistance instead of resistances alone. the midplane standoffs. unequally distributed : the 67 . to model at best the point where the midplane standoff is fastened. and a TMM.1 Nodal breakdown aluminum Frame In the Simplified Thermal model. each face of the aluminum frame was considered isothermal. As already pointed. the face is divided into nine nodes. modelling the conductive network between the nodes. the DTM consists of a GMM. near the face 3. 6. linking the aluminum frame with the COM PCB.1. two models have been developed. are connected at the top of faces 2 and 5. This is the aim of the present chapter presenting the Detailed Thermal Model (DTM). each face of the aluminum frame is now into several elements. It is divided into four main parts : developing. This was thus not taken into account in the STM. checking. For this reason. as the model detail evolves along with the design maturity of the CubeSat. used for the computation of the radiative exchange factors and environmental absorbed powers.Detailed Thermal Model 6 Introduction Up to this point.

Detailed Thermal Model Figure 6.2 – Detailed Thermal Model nodal breakdown of the aluminum frame Lionel Jacques 2nd Master in Aerospace Engineering 68 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 .1 – Midplane standoffs fastening to the aluminum frame Figure 6.Chapter 6.

The boundary between the corners and adjacent nodes is defined by the notch at the extremities of the folds of the faces 3 and 6. In practice. Lionel Jacques 2nd Master in Aerospace Engineering 69 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .1. based on these two criteria: . Equal elements imply easier computation of the conductive links within the panel but more difficulties for the transverse links with the frame as this will be exposed later. .COM : the only critical and high dissipation IC known at the moment is the amplifier. .OBC & OBC2 : few nodes required because of low power.2 shows the nodal breakdown of the frame. the connectors are modeled and added to the corresponding node as well as the spacers. 3605 and 3607 for the bottom face. CubeSat’s feets. where it is narrower. the dissipated power level is such that a finer discretization is also required ) 10 10. Another criterion was already spotted when considering the link with the secondary PCB sustaining the batteries : the transverse conduction within the PCB from main spacers at the corners has to be modeled. Matlab routines have been developed to make the discretization of each PCB entirely automatic and autonomous (both for the GMM and TMM) with only one input: the number of elements desired.EPS : more nodes required because of transverse conduction modeling and high power. Solar cells In the same way. the aluminum panels are as well divided but in nine equal elements. high dissipation ICs ) 10 10 elements such as the elements are about 1cm wide. This model takes into account the cell efficiency variation but this will be treated in the TMM. 3303. Nevertheless. PCBs Concerning the PCBs. approximatively the size of the dissipation transistor[51]. And here are the level of discretization of each PCB. even if it contains lower power ICs ) 10 10. Figure 6. In this model. more nodes are also required for transverse conduction modeling. aluminum panels To be consistent with the division of the frame. The complete numbering convention is available in appendix ??. 3305 and 3307 for the upper face and 3601. 3603. visible in figure 6. are also modelled and included in nodes 3301.Chapter 6. their level of discretization depends on their components: high localized dissipative components requires a finer discretization. . low dissipation ICs (integrated circuits) ) 2 2 elements.EPS2 : because the case of hanging the battery PCB on EPS2 will be studied. each solar cell has now its own node. Detailed Thermal Model nodes at the corners are smaller.

6. As shown by the lower left picture.7 (see previous chapter) is now used. Figure 6.05 ρd.1. machine finish) thermo-optical properties are set to = 0.8. it is only modelled with two nodes: one for the exterior and one for the internal side.3 [24]. The differences comes from the new elements taken into account : . Both antennas are 0.03 and α = 0.spacers (made of 6061 aluminum.1. the particular geometry of the frame has been modelled. Lionel Jacques 2nd Master in Aerospace Engineering 6. 6.2. . 6.antennas (copper) thermo-optical properties are: = 0.IR = 0. Since the antenna deployment mechanism panel geometry was not yet clearly defined when the model was created.IR = 0. the aluminum panels and frame radiative coupling are consistent with reality.the PCBs stack.Chapter 6. .2 Thermo-optical properties Thermo-optical properties are mainly the same as the ones used for the Simplified Thermal Model. .1 Conductive network 70 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .5 mm wide and are made of copper [63].2 Thermal Mathematical Model aluminum frame All the conductances of the frame have been determined through finite element analysis.3 ESARAD considerations The nodal breakdown defined above suggests us to split up the ESARAD geometry in 3 main parts : . antennas will probably be relatively hot. set as parameter : the effect of covering the batteries with a low emissive tape will be investigated. . Seeing their high α/ ratio.an external skin comprising the solar cells. the antennas and the remaining visible aluminum frame (mainly the rails and the feet).an internal skin representing the environment seen by the PCBs : the aluminum panels and frame. they are divided into several elements. Detailed Thermal Model Antennas This model includes the two antennas: the 17 cm and the 50 cm one. The detailed procedure is available in appendix B. .the emissivity of the rails : the measured value of 0. the aluminum panels.3 illustrates the GMM in ESARAD.25 mm thick and 3. In this way.15 ρs.the emissivity of the batteries. Once again.[50].

coming from Spacecraft Thermal Control Handbook by D.Chapter 6.57 mm).4 – TRW and Lockheed Martin Bolted-Joint Resistance Data [24] The fastening screws used are NC 4-40 (corresponding to M3 metric standard) and give.6 [K/W] contact resistance. shows contact resistance data1 for various bolt size and plate thickness. for thin plates (1.4. a 12. Detailed Thermal Model Figure 6. The table of figure 6. Data about screwed-joints are difficult to find because they are often developed within companies and seldom published.3 – Geometric Mathematical Model in ESARAD Measurements confirmed the presence of contact phenomenon between the lateral frame and faces 3 and 6. Figure 6. 1 from TRW and Lockheed Martin Lionel Jacques 2nd Master in Aerospace Engineering 71 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Gilmore.

But as the different layers are thin.Chapter 6.5 – Thermal conductance between two thin plates of different section In practice. Horizontal/vertical conductances (respectively R H and RV ) determination is then straightforward: w k (th) h k (tw) RH = RV = where w = Wpanels /3. h = H panels /3 and k the thermal conductivity of the 7075 aluminum panel. as the top and bottom panels (faces 3 & 6). equally distributed in both directions. the transverse conductance is calculated according to the figure 6. except for the antenna deployment mechanism panel. Therefore. mainly because the nodes are not directly opposed due to their different cross sections. 𝑆1 𝑡1 + 𝑡2 𝑆2 Figure 6. as the integration layouts (solar cells/panels and frame/panels) are the same for all the nodes. Lionel Jacques 2nd Master in Aerospace Engineering 72 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . 2 & 5) are identical. as defined above. the surface area involved in the thermal conductance formula (GL = kS ) is L taken as the smallest cross section of the two nodes. For example.5 where the heat flux flows from the upper plate to the lower one. neglecting their extension (and thus the increase of the local cross section) in the larger layer. the surface S would be equal to S1 and the length L = (t1 + t2 )/2 (the nodes are set in the middle of each plate thickness). Transverse links The thermal modelling of transverse links between the frame and the panels and between the panels and the solar cells is relatively complex. Detailed Thermal Model Aluminum panels There are nine nodes per panel. The three lateral panels (faces 1. and is computed through CATIA V5 CAD software. one can assume that the flux lines remains perpendicular to the cross section.6: 1 1 = Rij = GLij Se f f ective t panel tKapton tcell t t + adhesive + + adhesive + k cell k adhesive k Kapton k adhesive 2k panel for the solar cells/panels links and 1 1 = Rij = GLij Se f f ective t panel t f rame t + adhesive + 2k panel k adhesive 2k f rame for the frame/panels links where Se f f ective is the common area between nodes i and j. considering the case of the figure 6.

the PCBs are divided into several elements. Conduction through the endless screws is neglected (large clearance with spacers and washers) but their heat capacity is distributed on the 4 corresponding spacers. Figure 6. Detailed Thermal Model Solar cell 150𝜇𝑚 80𝜇𝑚 50𝜇𝑚 80𝜇𝑚 RTV S691 adhesive 2mil Kapton® RTV S691 adhesive 7075 Aluminium panel 1. both spacers and bus conductances are taken into account. The contact between the spacers and the washers is only ensured by the vertical pressure applied at the bottom of each endless screw. to take into account conduction inside the PCBs.7. k PCB t PC104 ly k L (t ) = PCB PC104 PC104 lx WPC104 k PCB (t PC104 lx ) k W (t ) GLy = = PCB PC104 PC104 ly L PC104 GL x = where lx = WPC104 and ly = LPC104 with N the number of elements in N N which the PCB is divided in both directions. Contact resistance is also added between the midplane standoff and the aluminum frame.27𝜇𝑚 or 1.52𝜇𝑚 Figure 6. Lionel Jacques 2nd Master in Aerospace Engineering 73 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .7 describes the inter PCBs conductive model. compared to the other resistances. Inner PCBs links Now. The copper being a good thermal conductor (k 400 [W/mK]).Chapter 6.6 – external layout and corresponding thermal conductance The adhesive thickness between the frame and the panels is not yet defined and the same values as the other adhesive layers are assumed. at the midplane standoffs.5𝑚𝑚 Adhesive ? ? 5052 Aluminium frame 1. The drill holes at the corners of the PCBs consist of copper washers for heat rejection and structural purposes as depicted in figure 6. Inter PCBs links Like in the STM. OBC PCB is supported by steel inserted threaded spacers into which endless screws are fastened. A contact resistance is thus added between the spacer and the PCB washer. conduction through the washer is neglected.

inserted copper washers in EPS2 engineering model. On the right.Chapter 6. Therefore. its geometry remaining unchanged. The default value corresponds to aluminum spacers. EPS PCB to battery PCB links The way the BAT PCB is fastened to the other PCB was considered as aluminum spacers in the STM. each node is linked to the previous and next node with the same conductance. the conductance between the BAT PCP and the EPS or EPS2 PCB is kept as a parameter. The antennas are also linked to the amplifier through a coaxial wire. the inter-PCBs conductive model Concerning the connector. The corresponding conductance consist of a 5 cm long copper wire with a 1 mm diameter for both antennas. only the pins contributes to the conduction and perfect contact is assumed between the pins of consecutive connectors. 6.2. STM also showed that the possibility of fastening the BAT PCB on the EPS2 instead of EPS should be investigated. The contact between both antennas and the panel is assumed to be perfect. On another hand. STM showed that insulating the BAT PBC by using spacers made of materials less conductive would be worthwhile. But before distributing the Lionel Jacques 2nd Master in Aerospace Engineering 74 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . The parameter who vary is the conductivity of the spacer.2 Internal power This section is devoted to the description of the internal power distribution which only concerns the hot case. with perfect contact. Detailed Thermal Model 𝑚𝑖𝑑𝑝𝑙𝑎𝑛𝑒 𝑠𝑡𝑎𝑛𝑑𝑜𝑓𝑓 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑤𝑎𝑠ℎ𝑒𝑟 𝐶𝑂𝑀 𝑃𝐶𝐵 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 ℎ𝑎𝑙𝑓 𝑠𝑝𝑎𝑐𝑒𝑟 4𝑡ℎ 𝑠𝑝𝑎𝑐𝑒𝑟 𝑖 + 1 ℎ𝑎𝑙𝑓 𝑠𝑝𝑎𝑐𝑒𝑟 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑃𝐶𝐵 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 ℎ𝑎𝑙𝑓 𝑠𝑝𝑎𝑐𝑒𝑟 𝑖𝑡ℎ 𝑠𝑝𝑎𝑐𝑒𝑟 𝑡ℎ Face 2 or 5 corner node 𝑠𝑝𝑎𝑐𝑒𝑟 3× 1𝑠𝑡 𝑠𝑝𝑎𝑐𝑒𝑟 ℎ𝑎𝑙𝑓 𝑠𝑝𝑎𝑐𝑒𝑟 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑂𝐵𝐶 𝑃𝐶𝐵 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑓𝑎𝑐𝑒 6 𝑠𝑡𝑒𝑒𝑙 𝑖𝑛𝑠𝑒𝑟𝑡 Face 6 corner node Figure 6.7 – On the left. Antennas Antennas being divided into several elements.

the fluence is 4.3E11 while over the predicted lifetime ( 4. over our mission duration of one year. Figure 6.28 2. data about the new solar cells are not yet available and the temperature coefficient is based on the 28% cells data.94 0. The efficiency is assumed to vary linearly with temperature1 according to the following law: ∂η ( T 28 C) ∂T where η0 is the nominal efficiency at 28 C.8 shows both effects of temperature and aging on the solar cells efficiency. However. since our solar cells are new ones with new aging coefficients still unknown and since they have not yet been tested in real space conditions the worst case value are used for the hot case. the coverglass transmittance decrease with the absorbed dose. The effect of deterioration is mostly visible on the nominal value of the efficiency but that the linear temperature dependence must surely be valid only in a given temperature range. 1 Note Lionel Jacques 2nd Master in Aerospace Engineering 75 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .1 : η ( T ) = η0 + ∂η ∂ = ∂T ∂T Pmax CS ASC = 1 CS ASC Vmax ∂Imax ∂Vmax + Imax ∂T ∂T where ASC is the cell area and CS the solar constant BOL Average efficiency η0 Voltage at Pmax . Spenvis simulations show that.1 0. However. For instance. Concerning dVMP /dT and dIMP /dT. Solar cells ef ciency The way the electrical power is subtracted has already been exposed in the STM. The column headers are the fluence of 1MeV electrons received by the cell expressed in [ Electrons/cm2 ].93 0. it must be collected: no power can be created out of nowhere and the section begins thus with the description of the solar cells.94 -6. this model takes into account the temperature dependence of the solar cells efficiency along with their degradation over time.36 0.8 years).1 – 28% solar cells efficiency parameters [7].97 0. summed up in table 6. the values does not decrease monotonically over time and their values in function of the fluence are directly available without the use of an aging coefficient.95 0. representing the deterioration over time. In those columns are the coefficients by which the efficiency.2 0.29 Table 6.3 -6. Detailed Thermal Model power.Chapter 6.5E14 5E14 1E15 aging factors [-] 0.86 0. VMP and IMP must be multiplied.4 0. VMP Current at Pmax . for CS = 1367 [W/m2 ] The last three columns represent the effect of aging due to the radiation environment: the cells efficiency deteriorates over time because the space radiation environment affects the materials used in the cells.99 0. it is 2E12.92 0.8 -6. IMP dVMP /dT dIMP /dT [%] [mV] [mA] [mV/ C] [mA/ C] 28 2371 487 -6.96 0. But as previously introduced.

5E14 EOL 5E14 EOL 1E15 0. The hot case is somewhat different: now that a more detailed power budget is available and that a more accurate distribution can be realized. all the collected power is dissipated through the transistor of the dissipation system. However only the more dissipative and less efficient components are modelled: the dissipation system and the amplifier. OBC and OBC2 have relatively low power consumption and only 2 0.34 0.36 BOL EOL 2.3 = 2. The position of both elements is not defined and so subjected to optimization. the batteries being fully charged. Assuming that the efficiency of the amplifier. two different hot cases are considered.75/0. the amplifier input power (0.5 = 625mW). According to the COM subsystem [36] and [28] and based on the link budget developed by MIAS [9]. The first one assumes that the CubeSat is nearly turned off and that.Chapter 6.75 W into heat.8 2.5/0. Since that time. At the time the STM was being developed. this means that a part of the input power of the converter is converted into heat (2.28 0. a fraction is dissipated into heat because the electronic components are not perfect and have their own efficiency. the higher its efficiency. the power required on the antennas is about 750 mW. The cold case assumptions are identical to the one of the STM: all the collected energy is stored under sunlight with no internal dissipation throughout the orbit.26 0.24 0.22 −60 −40 −20 0 20 Temperature [°C] 40 60 80 Figure 6. The second one assumes that D-STAR is turned on involving a full utilization of the amplifier. Following back the power path. not yet determined.5 W) is provided by the converter whose efficiency is about 80%.32 Cell efficiency η [%] 0. Once again. this means that the amplifier dissipates 1.05 Lionel Jacques 2nd Master in Aerospace Engineering 76 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . is at worst 30%.3 0. it has evolved and consists now in one transistor. the dissipation system was composed of four resistances dispatched at the corner of the EPS PCB.8 – Effect of temperature and aging on solar cells efficiency Internal power distribution Once the electrical power collected. Detailed Thermal Model the effect of temperature is clear : the cooler the solar cell. 0.

9 compares ESARAD MCRT and analytical results for one node of EPS2 PCB (16054) viewing COM PCB nodes (EPS2 and COM have the same number of nodes). uniform uniform uniform amplifier Table 6.3 Checking the model This section is devoted to the check of both GMM and TMM. Hot Case 1 0 0 Ptot 0 0 Hot Case 2 0.625 + Prem 0. the inovative EPS dissipation is about 0.B β ij = Fij j j + ∑ Fik (1 k k ) Bkj Details about the view factors computation is available in appendix C. in matrix form: B = β + (F β).3 W and the remaining power is finally dissipated through the dissipation system. The radiative coupling with the aluminum structure is modelled by closing the assembly with aluminum sides joining PCB perimeter. Gebhart factors Bij take into account multiple reflections and depend no longer only on the geometry like the view factors but also on the thermo-optical properties of the different surfaces.2 – DTM Hot Cases internal dissipation By default.18054 ). both amplifier and dissipation transistor are located at the center of their respective PCB 6. an analytical estimation of the REF between two PCBs is carried out using Gebhart’s theory and then compared to ESARAD results. The bus and the spacers are neglected. In addition to that. REF are directly obtained from the Gebhart factors Bij thanks to the following relation: GRij = i Ai Bij The analytical model includes two PCBs facing each other and divided in 10 10 elements. The main differences occur at the border Lionel Jacques 2nd Master in Aerospace Engineering 77 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .75 POBC [W] POBC2 [W] PEPS [W] PEPS2 [W] PCOM [W] diss. According to [33].Chapter 6.3 1.05 0.3. Figure 6. the number of rays fired by the MCRT method has been selected after a convergence analysis for different seeds but is not represented here. tansist.05 0. Detailed Thermal Model W are uniformly distributed on the two PCBs. The maximum REF is obviously with the opposite node (GR16054. 6. Gebhart factors Bij are defined by the equation: Bij = Fij or.1 GMM As for the STM.

Aluminum frame In order to check the lateral aluminum frame conduction. The comparison is presented in figure 6. Detailed Thermal Model because the analytical model does not take into account the connector. For this purpose. PCBs The PCB discretization is critical and therefore have to be tested.9 – ESARAD radiative coupling check for one PCB node (left: analytical. checking the conductive network seemed necessary. both surfaces are relatively identical and confirm that the number of rays fired is sufficient and that the REF computation through MCRT method has converged. Figure ?? shows the results of 1 For the sake of clarity. Lionel Jacques 2nd Master in Aerospace Engineering 78 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . one can verify the shape of the isotherms normally perpendicular to the heat flow. Figure 6. It confirms that the way the nodes are distributed and connected is consistent with the accurate temperature distribution of the FEM. the temperature is prescribed on face 4 while power is injected on the opposite face. the PCB inner links and the links between PCBs. The REF with the corners is indeed lower because the node does not see them directly and the corners have an area reduced by the presence of the spacer. The first one concerns the verification of the links between the nodes inside each PCB. The temperature reached is then compared with a finite element model under the same conditions. face 2 is undisplayed in ESARAD view. Apart these local differences. the gap between the PCB and the aluminum structure and the spacers.Chapter 6. right: MCRT) 6. being much more complex than the STM. three tests are carried out about different parts of the global conductive network: the lateral aluminum frame. Three checks are indeed carried out to verify the good implementation of the conductive network.3. By this way.101 .2 TMM Moreover. It is achieved by imposing a gradient across one diagonal of the PCB by prescribing the temperature of two opposite corners.

11 – Checking the conductive links inside PCBs through isotherm shape The second test concerns the links through the connectors. The way these links are tested is the following one: power is injected on EPS PCB while both OBC and COM have a prescribed temperature at their extremity. the links between OBC2-EPS). The temperature reached at the end of the EPS PCB. The shape of the isotherms are well as expected: nearly (because of the rectangular shape of the PCB) perpendicular to the diagonal of the PCB.10 – check this test. Detailed Thermal Model Figure 6. T2 . is obtained thanks to the following relation.12.Chapter 6. depicted in figure 6. The temperature reached where the power is injected is then compared to an simple conductive network. The automatic implementation is quite difficult because the number of elements of two consecutive PCBs may be different (for instance. deduced from the resistance model: T2 = T1 + Q R EPS + ( ROBC + Rs1 + Rs2 ) ( RCOM + Rs3 + Rs4 ) ROBC + Rs1 + Rs2 + RCOM + Rs3 + Rs4 79 University of Liège Applied Sciences Faculty Academic Year 2008 2009 Lionel Jacques 2nd Master in Aerospace Engineering . Figure 6.

2 W instead of 19. respectively. important differences will be observed in the temperatures thanks to the better accuracy of this thermal model.12 – Check of the conductive links through the bus and equivalent resistance model where R EPS = RCOM = ROBC = Rs1 = Rs2 = Rs4 = Rs2 = L PC104 n n 1 WBUS /2 k PCB t PC104 L PC104 n n 1 k PCB t PC104 15 mm k spacer Aspacer 25 mm k spacer Aspacer n = 10 n=2 For T1 = 0 C and Q = 5W one gets T2 = 261. 20.14 show the different contributions to the total power absorbed by the satellite during one orbit in the cold and hot cases.2 C while ESATAN gives 259 C. The global convergence of the TMM has been analyzed through the different convergence parameters () 6. Lionel Jacques 2nd Master in Aerospace Engineering 80 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . concrete simulations can be launched to analyze this initial design.4 Results Now that the DTM is checked.8 W in the hot case) is absorbed because of the antennas and feet contributions.Chapter 6.13 and 6.1 Absorbed powers Figures 6. 6. Nevertheless. They are nearly identical to those of the STM except that slightly more power (for instance.4. Detailed Thermal Model 𝑅𝐶𝑂𝑀 𝑅𝑠4 𝑅𝑠3 𝑅𝑠2 𝑅𝑠1 𝑇1 𝑅𝐸𝑃𝑆 𝑇2 𝑄 𝑅𝑂𝐵𝐶 𝑇1 Figure 6.

Chapter 6. Detailed Thermal Model 30 Solar Albedo Earth IR Total Total − Electrical Electrical 25 20 Absorbed power [W] 15 10 5 0 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 6.13 – Cold case absorbed power 25 Solar Albedo Earth IR Total Total − Electrical Electrical 20 Absorbed power [W] 15 10 5 0 0 10 20 30 40 50 60 Time [min] 70 80 90 100 Figure 6.14 – Hot case absorbed power Lionel Jacques 2nd Master in Aerospace Engineering 81 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

Chapter 6. Detailed Thermal Model 6. requiring much more cpu time.2 Cold Case Since the STM showed that the most critical components in the cold case were the batteries. this temperature is still too cold and heaters are required.3 Hot Case 1 The STM showed that a steady-state analysis using the mean absorbed powers would possibly be equivalent and much more time effective than a transient computation. Indeed. the two hot cases have also to be considered to keep a global point of view and allow a global design. This translates the effects of transverse conduction within the PCBs (here the EPS) and contact with spacers both increasing the total thermal resistance between the batteries and the structure. 10 5 0 −5 Temperature [°C] −10 −15 −20 −25 Solar Cells Face 1 Solar Cells Face 2 Solar Cells Face 3 Solar Cells Face 5 Solar Cells Face 6 Antenna deployment mechanism panel Battery 0 10 20 30 40 50 60 Time [min] 70 80 90 100 −30 −35 −40 Figure 6. as expected viewing the similarities in the absorbed powers of STM and DTM.4. they are the only internal part represented in Figure ?? showing also the temperature evolution and solar cells. a comparison with transient analysis confirmed that the steady-state one was equivalent and that the fast variations of the absorbed powers does not influence the mean results. But before beginning the design of the heater. Lionel Jacques 2nd Master in Aerospace Engineering 82 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . it drops now only to -15 C. 6.4. steady-state analysis are then performed for the hot case instead of transient.15 – Cold case absorbed power Yet. The first key difference with the STM is clearly visible: if one reminds that the temperature of the batteries dropped to -21 C in the STM. From now.

Lionel Jacques 2nd Master in Aerospace Engineering 83 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .56 W 3.65 W 3.1°C) EPS2 (39.15 W 0.4°C) 1400 mW Spacers : 75 % Bus : 25 % EPS1 (55.The solar cells are the most exchanging surfaces rejecting nearly as much power as they receive (about 12 W).36 W 0. While the heat flow path are not critical in the cold case because there are no internal dissipation and because the structure does not undergo large thermal gradients. Figure 6.43 W Aluminum panels (32. Each of the twelve components represented in this diagram involves several underlying nodes so that temperatures and heat flows are averaged.76 W 0.3°C) 1630 mW Spacers : 57 % Bus : 43 % OBC2 (39.The links joining the PCBs gives also the distribution of the heat flow between the spacers and the connector.16 summarized the results of the first hot case. .3°C) 1140 mW 1180 mW Spacers : 29 % Spacers : 61 % Bus : 71 % Bus : 39 % 3.11 W 0.89 W = 3.61 W 0.99 W P 3.68 W 2. All the external parts of the CubeSat are at the top while the PCB stack is at the bottom.56 W Legend : Radiative : Conductive : Incident power : Evacuated power : Dissipated power Figure 6. a heat flow map is more useful in the hot case.1°C) 40 mW BAT PCB (48°C) 0 mW 1460 mW face 2 : 48% face 5 : 52% 20 mW COM (36.7°C) 0.1°C) 170 mW Aluminum frame (32. 12.17 W 3. The three red arrows represent the total effective power exchanged radiatively between the PCBs and the external parts.52 W Solar cells (32°C) 3.Chapter 6. encircled by the red dotted rectangle. Therefore.03 W 11. we created a Matlab routine that generates automatically the heat flow map from ESATAN output file to synthesized the results.2. Detailed Thermal Model As described in the table 6.23 W elec Antennas (126.5°C) Batteries (48.8°C) 30 mW 1740 mW (face 3) OBC1 (36. This highlights that both are as important as the other.16 – Hot case 1 heat ow map This diagram leads to the following observations: .3°C) 170 mW 3200 mW Antenna deployment mechanism panel (32. the first hot case assumes that all the collected power is dissipated through the dedicated transistor. The different radiative exchanges occurring between the PCBs are not represented but can be estimated by differentiating the power incoming and leaving conductively one PCB being approximatively equal to the power evacuated radiatively by the PCB.

17 shows that the temperature near the transistor rises up to 115 C. it has approximatively the same temperature as the first hot case Lionel Jacques 2nd Master in Aerospace Engineering 84 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 .Chapter 6. . Concerning the structure. mainly because the hot case is different (in the STM.22 W). . .The mean electrical power collected is more than 3. Detailed Thermal Model Figure 6.17 (right) details the temperature distribution on the antennas and shows that the maximum reached temperature occurs at the extremity: about 200 C.4. the dissipated power was simply divided between the COM and the EPS because a more accurate distribution was not available) .The global radiative exchange between the PCB stack and the structure is small compared to conductive one: about 10% (0.17 – Hot case 1: Hot spot due to the dissipation transistor and temperature distribution on the antennas .37 W / 3. It is worst than that was predicted by the STM.The batteries are out of their allowed range of temperatures (in charge): 48 C.Antennas are quite hot: their mean temperature is about 125 C but the corresponding panel remains cold. 6. one observes that OBC.5 W and is directly and entirely dissipated on the EPS.The conductive heat flux between EPS and BAT PCBs is null on average because the four spacers balance each other.18). OBC2 and EPS PCBs are cooler on average while EPS2 and COM are hotter (Figure 6. 20 mW flow through the coaxial cable to the amplifier. Figure 6.The total conductive heat flux flowing through the bus and spacers increases from EPS PCB to the outer PCBs because the radiative heat flux between to consecutive PCBs is added at each step. .Hot spots are not visible on this diagram but the mean temperature of the EPS PCB is 55 C while other PCB’s temperature ranges from 35 to 40 C.4 Hot Case 2 When analyzing the second hot case where the dissipated power is less localized. . The left picture of Figure 6.

6°C) BAT PCB (41.75W dissipated power.56 W 3. the amplifier is located at the center of the PCB. With its efficiency of only 30% involving a 1.9°C) 3. This is due to the amplifier and the coaxial cable.3°C) 0.19 shows the local hot spots on EPS and COM PCBs. The batteries are cooler than in the first case but yet remain unacceptably close to the 45 C allowed temperature in charge.68 W 2.1°C) 160 mW Spacers : 165 % Bus : −65 % 300 mW : Incident power : Evacuated power : Dissipated power Figure 6. 6.3°C) 170 mW 2880 mW Antenna deployment mechanism panel (33.41 W 0. By default.41 W EPS2 (40. These local effects could not be revealed in the precedent thermal model. like the antennas. for instance between EPS1 and EPS2 where 160mW flows effectively from EPS1 to EPS2 but 264mW through the spacers and 104mW through the connector in the opposite direction.7°C) Batteries (41.76 W 0.61 W 0. Figure 6. Detailed Thermal Model except for the antennas panel being a little hotter.23 W Solar cells (31.52 W 11. Once again. the temperature of the amplifier reaches 70 C.Chapter 6.98 W Pelec = 3.15 W 0. The four issues for which solutions have to be found are namely: Lionel Jacques 2nd Master in Aerospace Engineering 85 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .02 W Antennas (130.1°C) 20 mW 1750 mW 1710 mW face 2 : 53% face 5 : 47% 300 mW COM (44°C) 610 mW Spacers : 121 % Bus : −21 % 1060 mW Spacers : 55 % Bus : 45 % OBC2 (36. These are approximate numbers because the coaxial cable is assumed to be directly connected to the amplifier.17 W 3.18 – Hot case 2 heat ow map Little loops in the heat flow also appears.65 W 3.44 W Aluminium panels (32°C) 170 mW Aluminium frame (32. this temperature is relatively close to the upper limit of the industrial temperatures range (+85 C) and must be carefully looked at. 12.1°C) 20 mW EPS1 (43°C) 870 mW Spacers : 31 % Bus : 69 % 50 mW Legend : Radiative : Conductive 1.5 Parametric analysis and design Previous results have confirmed two problems already spotted thanks to the STM but revealed two others.4°C) 20 mW 1170 mW (face 3) 50 mW OBC1 (34. through which 300mW (of the 1750mW dissipated by the transistor) flows.88 W 3.68 W 0.

1 Cold case issues The main problem of the cold case is the temperature of the batteries dropping to -15 C when the satellite comes out of the eclipse. In the hot case. the main problem is not the transistor itself because it is designed to support such temperatures but the components close to it.5 W power dissipation [4]). the temperature of the batteries is too high. Lionel Jacques 2nd Master in Aerospace Engineering 86 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 . Detailed Thermal Model Figure 6. A efficient way of heating the batteries will thus be investigated in this section. 4. According to its datasheet. involving the narrower range of temperatures (0 +45 C). Nevertheless. time at which the batteries begin to charge again. This problem will thus not be plainly investigated in this work but should be as soon as more information about the amplifier are available.Chapter 6. The second issue concerns the dissipation transistor. avoiding hot spots and reducing the temperature of the EPS PCB would be favorable for the transistor as well as for the batteries.5. the amplifier reaches too high temperature. A way to enhance heat rejection must be found to reduce the temperature reached by the batteries 3. 2.19 – Hot case 2: Hot spots due to the dissipation transistor and amplifier on COM PCB 1.e. Nevertheless. 6. concrete measures should not be taken before a better definition of the amplifier because they could be inadequate if it was effectively more efficient. as the amplifier is not yet determined and could be much more efficient (like the ADL5541 linear amplifier which has a constant 0. In the second hot case. In the cold case. the temperature of the batteries drops below its acceptable lower limit at the end of the eclipse. This is thus the lower limit of the charge temperature range that must be considered i. 0 C. An this is indeed the more critical time: when they come into sunlight and get recharged. when maximal dissipation occurs.

two options are possible. and this has been confirmed by MIAS simulations. we preferred the second solution: a too cold battery is useless even if it has a low voltage and it is more reliable in case of OBC failure (the heating system is still operational and the batteries can still feed the beacon). Lionel Jacques 2nd Master in Aerospace Engineering 87 University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 . EPS [55].20 illustrates the decreasing efficiency and capacity of a typical LiPo battery (here the KOKAM SLPB723870H4) with the temperature. Figure 6.20 – Battery temperature profile of a typical LiPo battery. The decision of whether it is wise to heat the batteries when they are already nearly discharged is then more complex. The second way of control consist in bypassing the OBC and develop an independent and autonomous control loop. Detailed Thermal Model The use of a heater seems the easiest and more convenient way to prevent the temperature from dropping below this limit. Nevertheless. After many discussions with the other subsystems and other CubeSat teams. The -10 C and 0 C curves illustrate than a heating system is well required. Then what is the optimal threshold below which it is turned on? The power generated by the heater is not entirely accumulated in the batteries: there are losses which does not participate to the heating of the batteries. Figure 6. several questions arise: How perform the control of the heater? What is the required power? Because the response is obviously not instantaneous. here the KOKAM SLPB723870H4 [32] Moreover. The first one involves the On-Board Computer (OBC) and allow to take into account other criteria in addition to the temperature of the batteries such as their voltage.Chapter 6. the batteries will keep getting cooler a short time after the heater is switched on. How to reduce these losses and thus increase the heater efficiency? Concerning the way to control the heater. when designing a heating system. EPS2 [33] designed the batteries in such a way that their capacity should not drop below 70%.

the other is still operational. the heater is switched on until the batteries are again within its allowable temperature range. Actually.03) is low outgassing and would be suitable. two corresponding ways of insulating the batteries have been foreseen and investigated: covering the batteries with an aluminum tape (low emissivity) to reduce radiative losses.21 shows the resulting control loop including the sensor and the heater: if the temperature of the batteries drops under a given threshold. Furthermore. fastened to another PCB). As early presented in this work. and using insulating materials for the fasteners to reduce conduction losses.21 – Heating system control ow chart As the batteries are located on either side of the BAT PCB. Sheldahl’s first surface aluminized polyimide tape with 966 acrylic adhesive has a low emissivity ( 0. in agreement with STRU subsystem. only two ways of heat transfer occur in space: radiation and conduction (except when considering aerothermal flux or specific thermal control means such as fluid loops or phase-change materials). Detailed Thermal Model Figure 6. Fully replacing the initial aluminum spacers by a thermoplastic materials must then be avoided because the BAT PCB undergoes strong dynamic amplifications (being in the center of the CubeSat. On another hand. the third one enables the defective sensor determination. Diminishing the losses is achieved by insulating the batteries and their PCB. usual insulating materials are thermoplastics such as PTFE (Teflon R ) or PA (polyamide such as nylon) but these have poor structural properties and outgassing must be carefully looked at. From this observation. we decided to use two separated redundant heaters in parallel with two sensors to increase the reliability of the system. Aluminum tape is easy to use and largely used in space applications. here is the outcoming design depicted in Figure 6. Concerning the BAT PCB fasteners. To answer the three other questions about the required power. using two identical heaters in parallel ensures that if one fails. Structural integrity is enforced by titanium spacers and screws (high tensile strength Lionel Jacques 2nd Master in Aerospace Engineering 88 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .22. structural integrity could be weakened because of the creep phenomenon. a third sensor should be used: if two sensors give different temperatures. Temperature sensor Heater on 𝑇 > 𝑇𝑇ℎ𝑟 ? Yes No Heater off Figure 6. For those reasons.Chapter 6.33 [52]. parametric analyses are performed. threshold and losses reduction. According the the Sheldahl Red Book p.

The right plot shows that for thick washers (above 1 mm). was 73. CES Edupack value) reduce the equivalent thermal conductivity from 3. According to CES software. After implementing the heating system in ESATAN. Detailed Thermal Model and Young’s Modulus to thermal conductivity ratio) in addition to thin nylon washers. Figure 6.24 shows the minimum temperature reached on the batteries in function of the heater power for different insulating configuration.5 mm thick washers replacing aluminum (k = 150 W/mK) by titanium (k = 10 W/mK. considering 0. the insulation benefit is quantified. the thermal conductivity of PA (nylon 6. Based on this value.23 shows the influence of the washer thickness and thermal conductivity of the spacer. The conductance is thus divided by a 55 factor. reducing creep issues and increasing thermal insulation.22 – Batteries’ PCB fastening concept The resulting insulation gain can easily be evaluated in considering only the main heat transfer through the spacer (conduction through the screw is neglected because of poor contact around the nut) and a perfect contact between the PCBs and washers. The equivalent resistance compared to an aluminum spacer of the same dimensions is far much greater: Lspacer k alu Aspacer Lspacer 2 twasher 2t + washer k titanium k washer 1 Aspacer Lspacer k eq Aspacer R alu = & Req = = This defines the equivalent thermal conductivity which will be used as parameter: k eq = k washer k titanium k washer (1 γ) + k titanium γ washer where γ = 2Ltspacer .32 10 3 W/K while the initial one. their concrete effects are studied. The effect of radiative insulation is clearly observed: for a same equivalent thermal conductivity Lionel Jacques 2nd Master in Aerospace Engineering 89 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Nylon has been preferred because of its higher Young’s Modulus.e. i. the Figure 6. The left plot demonstrates that 0. Eventually. The threshold is set to 5 C and the heater power is varied (the total heater power for the two batteries).5 mm thick PA washer combined with titanium spacer. For instance. considering 0. with the aluminum spacer. the global thermal conductance is now equal to 1. the thermal conductivity of the spacer is less influent than for thinner washers.5 to 2.Chapter 6. Now that solution have been proposed. a 23% gain.6) is 0. Titanium screw BAT PCB Titanium spacer washers EPS PCB Figure 6.5 to 1 mm thick washers already reduce efficiently the conduction and using thicker washers would be useless.7 W/mK.24 W/mK.5 10 3 W/K.

plastic) shows that a more than 500 mW heater is required to reach the threshold. reducing the emissivity of the batteries through aluminum Lionel Jacques 2nd Master in Aerospace Engineering 90 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . k=100 0 −5 −10 −15 100 200 300 400 500 600 Heater Power [mW] 700 800 900 Figure 6.5 mm washer 1 mm washer 2 mm washer 5 mm washer 150 Figure 6.23 – Equivalent conductivity of BAT PCB fastening in function of the PA washer thickness and thermal conductivity of the spacer 10 high εbattery.e. On the other hand. k=2 low εbattery. Detailed Thermal Model 20 16 14 equivalent thermal conductivity [W/mK] equivalent thermal conductivity [W/mK] 12 10 8 6 4 2 18 16 14 12 10 8 6 4 2 100 50 5 spacer thermal conductivity [W/mK] 4 3 2 1 0 0 washer thikness [mm] 50 100 spacer thermal conductivity [W/mK] 150 0. k=2 low εbattery.Chapter 6. Threshold = 5°C 5 low εbattery.24 – Effect of insulating the batteries on the heater required power for a given threshold [5 C] (2 W/mK) the blue doted line corresponding to the high emissivity (the initial value considered for the batteries was 0.1 mm washer 0. k=10 Battery Minimal Temperature [°C].2 mm washer 0.8 i.

The 40 mAh consumption that could be saved by reducing the threshold remains quite C= 1 The time the heater is on is obtained through the post-processing of the ESATAN output file.24 also shows that increasing the insulation of the batteries increases the heater sensibility ( dTbatteries ): the slope of the 2 W/mK curve (blue) is greater than the slope dPheater of the 10 W/mK or 100 W/mK one (before the threshold is reached).5 V has been assumed to be conservative.03) lowers the required power to 300 mW. Detailed Thermal Model tape ( = 0.Chapter 6. The power generated by the heater is thus multiplied by the time during which it is turned on and then divided by the voltage. Finally. the time it is turned on can be reduced to 35 to 50 min. Three parameters are looked at in this study: . the low the require power to reach it. The worst case corresponds therefore to the minimum voltage: a 2.20 showed that the maximum voltage is 4.25 shows the evolution of the three parameters presented above in function of the power and for different thresholds.and the consumption of the heating system. Bottom plot shows that reducing the threshold and increasing the power reduces the time the heater is turned on.the minimal temperature reached by the batteries to verify that the power is sufficient.7 V. . Nevertheless. Figure 6. Not reducing the emissivity of the batteries is nearly equivalent to increase the equivalent thermal conductivity to 10 W/mK as shows the red curve. Top plot confirms that the lower the threshold.20 suggest therefore that sufficient safety margins to cope with these uncertainties combined to the model and environmental uncertainties.the time during which the heater is turned on.2 V while cut-off voltage is 2. the thermal sensors and heater have a finite accuracy and LiPo battery’s characteristic curves illustrated in Figure 6. Figure 6. Pt [mAh] U The heater will be directly connected on the batteries to enhance reliability. The remaining two questions concern the required power and the switching threshold. The voltage is therefore not constant because it is not converted by one of the EPS converter. set here to 360 points per orbit Lionel Jacques 2nd Master in Aerospace Engineering 91 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . For a 500 mW heater. The consumption of the heating system is computed as follows: the energy of the battery is given by its capacity. the black curve demonstrate that even a 1W heater would not be sufficient if the conductive insulation is not achieved. . i. given in mAh. 35 to 50% of the orbital period (104 min) while the maximum eclipse time is about 35 % of the orbital period. Figure 6. Middle plot also shows that reducing the threshold from 5 to 0 C allows to save 40 mAh. The time the heater is turned on consequently depends on the output time step.e. The middle plot shows that the increase power resulting in a reduced consumption time balance each other once the threshold is reached: the capacity curves are approximatively horizontal (due to computation error of the consumption time1 ) after that point. Two of the four initial questions have been answered. depending on the threshold.

A persistent deformation has been noticed on both models so that batteries should be packed.25 – Evolution of the battery minimal temperature. recalling that the worst case has been assumed with the 2.5 V voltage. The same phenomenon has been observed for the VARTA PoLiFlex R battery. kBAT=1W/mK 400 450 500 150 200 250 300 350 400 450 500 Figure 6. the threshold shall be set to 5 C and we shall use a 500 mW heater. Battery Minimal Temperature [°C] 5 Threshold = 0°C Threshold = 1°C Threshold = 2°C Threshold = 3°C Threshold = 4°C Threshold = 5°C 150 200 250 300 350 400 450 500 0 −5 100 Capacity [mAh]. Figure 6. batteries’ capacity decrease (worst voltage U=2. Nevertheless. the determination of the generated power is impossible through this data because this was not the purpose of this test. instead of using one heater for the two batteries. even if this was not the subject of these tests. vacuum tests have been conducted by EPS subsystem on different batteries (KOKAM and Varta) to verify their functionality in their allowable temperature range. Detailed Thermal Model small compared to the nominal 1000 to 1500 mAh capacity of the foreseen batteries. U=2. Figure 6. Moreover. Results are available in the test report [34]. The 500 mW have to be divided into two 250 mW heaters.5 V) and time the heater is turned on in function of the heater power for different thresholds For the sake of reliability. In addition to that. the remaining 250 mW are able to ensure that the batteries’ temperature will not go below 0 C. For the same reason.Chapter 6. each battery shall have its own heater as previously stated.26 illustrates increase of temperature: about 5 C whereas the batteries are fastened to a temperature controlled panel through a nitrogen loop. heating has also been noticed under discharge.25’s top plot shows that even if one fails. low εbattery. the conclusion is that the batteries undergo a self heating during discharge Lionel Jacques 2nd Master in Aerospace Engineering Time the heater is on [min] 92 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .5V 160 140 120 100 80 60 100 120 100 80 60 40 20 100 150 200 250 300 350 Power [mW]. However.

However. it is difficult to determine the heater model. based on the typical dimensions of the different foreseen models.2°C temperature increase because of discharge Time Figure 6. Temperature decrease before discharge End of discharge and temperature decrease T° [K] Beginning of discharge 5. Lionel Jacques 2nd Master in Aerospace Engineering 93 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .R26. Detailed Thermal Model which is favorable: when supplying the heaters.Chapter 6.3 specifies the resistance value : 26. we found the following MINCO reference [40]: XHK5377R26.5377 specifies for the model size and shape: 35.B specifies Pressure-sensitive adhesive (PSA) If it appears that the batteries should be packed in a thin aluminum box with epoxy resin.4 mm rectan- .3 ω (This was the closest value from the one defined above corresponding to the dimensions of the batteries) . additional heat is automatically generated inside the batteries.52 = = 25 [Ω] P 0.HK stands for polyimide heater . adhesive could possibly be removed (XHK5377R26.3L12B where .X means low outgassing ink . Using Ohm’s law.3L12A).L12 is the lead length in inches (12 inches is the standard length. .6 gular shape 59. the electrical resistance is equal to R= U2 2. other length are possible on demand).25 As the battery model is still unknown.26 – Temperature of a SLB 603870H KOKAM battery during the cold test [34] To determine the resistance of the heaters. the lowest voltage is once again assumed.

2 Hot case issues In order to deal with the hot case issues. involving a equivalent resistance of 2. Because this resistance value was not available among suitable shapes. three solutions have been investigated: 1. as already introduced. we decided to use two 4.18 and 6.27.7 Ω resistances in parallel. the second possible solution investigated. the power distribution between the equivalent resistance and the transistor is depicted in Figure 6. The bottom plot shows that the power dissipated by transistor is already divided by two for the 3. 5 4 Power [W] 3 2 1 0 Resistances Transistor 0 1 2 3 Total dissipated power [W] 4 5 6 Percentage of total dissipated power [%] 100 80 60 40 20 0 0 1 2 3 Total dissipated power [W] 4 5 6 Figure 6. as introduced above for the cold case issues. In agreement with the EPS subsystem. is also investigated. The effect of insulating batteries’ PCB. The EPS subsystem determined that the ideal resistance to prevent the transistor from depolarizing was 2.35 Ω. With this configuration and according to the EPS subsystem. we suggested to relocate a part of the power initially dissipated through the transistor by adding a resistance in series with the transistor and fastening it onto the antenna deployment mechanism panel.Chapter 6. Detailed Thermal Model 6. is to fasten the batteries’ PCB to the EPS2 PCB instead of the EPS one because the EPS2 PCB is less dissipative and cooler. For this purpose.16. TM the use of an adhesive Minco Thermofoil heater [40] seamed convenient thanks to their easy integration and wide range of shapes.5. To reduce the temperature of the batteries.4 Ohms.5 W mean collected power computed and shown in Figures 6.27 – Power distribution in the equivalent resistance and the transistor of the dissipation system 2. Two equivalent thermal conductivities are considered: 150 W/mK (full aluminum spacers) and 2 W/mK (combination of titanium screws Lionel Jacques 2nd Master in Aerospace Engineering 94 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

The effect of the equivalent thermal conductivity starts to be significant when the BAT PCB is fastened to EPS2 Lionel Jacques 2nd Master in Aerospace Engineering 95 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . We designed a copper angle bracket that would be strongly bolted on the EPS PCB with two M3 bolts. in agreement with STRU and MECH subsystems. the first half gives the temperature for the different combinations without strap. The third solution is to use a thermal strap between the transistor and the antenna deployment mechanism panel. The first row recalls the initial design (first two columns) and shows the effect of the relocated power without any other change. As radiative insulation has proved to be useful in the cold case. The different combinations are compared through the batteries and transistor temperatures. Figure 6. Detailed Thermal Model and spacers with Nylon or PTFE washers). low emissivity is assumed (batteries covered with aluminum tape). The transistor temperature is reduced by a 1. Resistances in parallel Thermal strap (Angle bracket) Figure 6. The next eight rows assumes that the batteries are covered with aluminum tape as it has proved to be useful when examining the cold case. A combination is described through the first four columns in which are given the PCB to which the BAT PCB is fastened (EPS/EPS2) and the fastening equivalent thermal conductivity (k BAT ) the strap column shows whether the transistor is strapped to the antennas’ panel or not.28 shows the angle bracket in red encircling the transistor (left) on EPS PCB engineering model and a CATIA model illustrating the two adhesive resistances and angle bracket configuration.3 for the different combinations.5 factor while a 7 C battery temperature decrease is already achieved. The effect of relocating a part of the power through the two Minco resistances is given in the last two columns.28 – Thermal strap (angle bracket in red) design on engineering EPS PCB (left) and its fastening configuration with the relocated resistances on antenna panel (CATIA Model on the right) The effects of these three solutions are summarized in Table 6. 3. Among these eight rows.Chapter 6. encircling the transistor. then located close to the face 4.

black paint enables another 1 C decrease. On OUFTI-1. This can be easily explained: the heat radiatively received by the batteries must be evacuated to reduce their temperature.3 – Effects of the different solutions to the hot case issues To make the temperature more uniform inside spacecraft. A slight increase of the transistor temperature is also noticed. less than 60% of its initial value.e. Detailed Thermal Model and a higher thermal conductivity would be preferred if it should not be avoided for the reasons exposed in the cold case analysis. The second half of the table shows the results when the thermal strap is added and demonstrates that the combination of the strap with the Minco resistances can reduce the temperature of the batteries to 35 C and the temperature of the transistor to 67 C. The emissivity of the 6 aluminum panels has therefore been set to 0. A black coating of the aluminum panels implies therefore not enough significant changes in the temperature to be worthwhile. strapping the transistor to the antennas’ panel.Chapter 6. only the aluminum panels can be coated with a black paint because Pumpkin Kit’s Structure is not intended to be modified. without resistances k BAT Tbatteries Ttransistor [W/mK] [ C] [ C] 150 48 114 150 48 115 2 47 115 150 39 115 2 45 115 150 38 94 2 37 95 150 35 95 2 36 95 high aluminum panel (0. reliability and uncertainties suggest to combine both solutions. black paint is usually used on internal structure elements to increase their emissivity (being low because spacecraft’s structure is usually made of aluminum). Figure 6. In the last combination. the nominal case with batteries covered with aluminum tape and the final design.e.3 for three combinations: the nominal case.8 (inside) and the results are available in the last three row of Table 6.8) 150 46 112 150 47 113 150 35 94 with resistances Tbatteries Ttransistor [ C] [ C] 41 75 41 76 40 77 36 77 39 77 37 67 35 67 34 67 35 67 43 43 34 80 80 66 strap high bat EPS/EPS2 EPS EPS EPS EPS2 EPS2 EPS EPS EPS2 EPS2 EPS EPS EPS low bat no no no no no yes yes yes yes no no no high low bat bat Table 6. relocating of a part of the dissipation system power and insulating the Lionel Jacques 2nd Master in Aerospace Engineering 96 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . High emissivity on both aluminum panels and batteries allow a 2 C decrease of the batteries and transistor temperatures in the nominal design and a 1 C decrease if the batteries are covered with aluminum tape.29 shows the heat flow map resulting from the modifications discussed above i. i. Even if fastening the batteries’ PCB to EPS2 does not appear to be anymore worthwhile when strapping the transistor. Finally.

67 W 2.2°C) 360 mW 250 mW Spacers : 75 % Spacers : 89 % Bus : 25 % Bus : 11 % 1. Radiation is neglected and the PCB is assumed to be an insulated plate subjected to some boundary conditions and internal heat loads.2°C) 0.17 W 3.9°C) 50 mW Aluminum frame (32.3 W 0. Detailed Thermal Model batteries with aluminum tape.52 W 11. this kind of problems can be reduced to a single matrix equation: AT=g where T is the 1 n vector containing the n temperatures of the discretized problem. Table 6.Chapter 6.95 W P 3.56 W 3.87 W Legend : Radiative : Conductive : Incident power : Evacuated power : Dissipated power Figure 6.46 W Aluminum panels (31. For the sake of clarity. 12.8°C) 3. It shows that nearly one half of the power is relocated and that 1 W of the 1.78 W remaining dissipated power on the transistor is directly evacuated through the strap and less than 1 W has still to be evacuated through the spacers. no design modifications were investigated.9°C) 1090 mW 350 mW face 2 : 59% face 5 : 41% 80 mW COM (33.23 W Solar cells (31. the underlying equations are not detailed here but are available in appendix E. The power incident to the amplifier is however slightly increased because the antennas’ panel is hotter (80 mW instead of 20 mW)) but this has no significant effect on the temperature of the amplifier.9°C) 0 mW 0 mW BAT PCB (34. the parametric analysis has been carried out through some simplifications.15 W 0.6°C) 40 mW 790 mW Antenna deployment mechanism panel (37. We only studied the effect of the location of the amplifier on the PCB and its dissipated power.99 W elec Antennas (129.2°C) 400 mW Spacers : 73 % Bus : 27 % OBC2 (34. g the heat loads incident on each node and A is a n n matrix represents the conductive coupling between the nodes. because too high uncertainties remains about the dissipated power. We also verified that EPS2 fastening has no significant effect on the heating system. To avoid numerous costly ESATAN simulations.9°C) 280 mW Spacers : 91 % Bus : 9 % EPS1 (36. Using the finite difference method.4 summarized the achieved temperatures.75 W 0.78 W 2.97 W = 3.7°C) EPS2 (34.2°C) Batteries (34.69 W 440 mW (face 3) OBC1 (33.65 W 2.29 – New design hot case heat ow map Concerning the COM amplifier. Lionel Jacques 2nd Master in Aerospace Engineering 97 University of Liège Applied Sciences Faculty Academic Year 2008 2009 .61 W 0.5°C) 10 mW 1.

6 0. The effect of local heat dissipation has been studied in function of its location on the PCB. But getting closer to the corners reduces the temperature of the amplifier.30 – Amplifier temperature increase in function of its relative location on the COM PCB Based on this model. Figure 6.8 1 of the hot relative position spot in x direc tion Figure 6.Chapter 6. It is worthwhile to notice that setting the amplifier at the center of the PCB is not the worst situation (50 C temperature increase in comparison to the Lionel Jacques 2nd Master in Aerospace Engineering 98 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Detailed Thermal Model Element Solar cells Aluminum panels Aluminum frame Antennas’ panel Antennas OBC PCB OBC2 PCB EPS PCB EPS2 PCB COM PCB Batteries Dissipation Transistor T before [ C] 32 32 32 33 127 36 40 55 39 36 48 114 T after [ C] 32 32 33 38 129 33 34 37 34 34 35 67 Table 6.4 0. The dissipated power is maintained at 1.2 0. the COM PCB has been discretized and the temperatures of its four corners are supposed to be constant which is not far from reality. The worst locations involving the greater increase of temperature are the center of the four edges of the PCB.75 W.5 pot in yd irec tion 0 0 0.4 – Comparison of hot case temperatures before and after modifications 70 60 relative temperature increase [°C] 50 40 30 20 10 rela tive pos 0 1 itio no f th eh ot s 0.30 shows the temperature increase on the amplifier in function of its relative position on the PCB.

The four issues that have been confirmed or revealed are namely: . The resulting temperature increase drops down to only 15 C at the center of the PCB (and the reached temperature given by ESATAN is about 45 C). One must bear in mind that the bus has been neglected as well as the radiative exchanges. Insulating the batteries has proved to be required to reduce losses and consequently increase the heating system efficiency. For instance. the present model enable new issues detection.19). which shall be set to 5 C because of sensor and comparator finite accuracy. The foreseen Minco heaters has not yet been ordered because the definitive battery model is not defined. In addition to that.the too high temperature reached on the amplifier. far much complicated and computationtime costly than the previous one. Two 250 mW heaters in parallel. . This involves another 3 C and 30 C temperature decrease of the batteries and transistor. comprehensive tests should be conducted to validate the design a evaluate the sensors and comparator accuracy. Bearing in mind the problems encountered by Compass-1. even if the batteries are not defined. one for each battery. are required to maintain the temperature above the threshold. The use of another amplifier such as the ADL5541 linear amplifier which dissipates a constant 0.e. 75 C while the temperature of the corners is about 30 C. a part of the dissipated power is now relocated through two resistances in parallel also fastened to the antennas’ panel. which explains the difference. Two main measures have been investigated to reduce both batteries and dissipation transistor temperatures: an angle bracket encircling the transistor and fastened to antennas’ deployment mechanism panel already allows a 10 C temperature decrease of the batteries (from 48 C to 38 C) and 20 C of the transistor (from 115 C to 95 C). Detailed Thermal Model corner’s temperature) and that this corresponds indeed to the temperature observed in ESATAN results (Figure 6.their too high temperature in the hot case. using the ADL5541 could reduce the temperature down to 45 C. Concrete measures have been taken to solve three of these four issues. we stated that the efficiency of the amplifier should be taken into account when selecting the amplifier. Only three because too many uncertainties still surround the COM amplifier. However. i.5 W power [4] has also been investigated. respectively Lionel Jacques 2nd Master in Aerospace Engineering 99 University of Liège Applied Sciences Faculty Academic Year 2008 2009 . has however proved worthwhile.the hot spot due to the dissipation transistor. However. . a 45 C increase. A heating system has been designed to cope with batteries’ too cold temperature.Chapter 6. the problems were involved by the dissipation system.the too cold batteries in the cold case. Concerning the hot case. especially in the hot case. The issues already revealed by the Simplified Thermal Model became more tangible and evident. . On another hand. we suggest to use three temperature sensors for reliability purposes. Summary The Detailed Thermal Model.

tests should be conducted to evaluate the real efficiency of these measures. Detailed Thermal Model reduced to 35 C and 67 C. Lionel Jacques 100 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Once again.Chapter 6.

the different test and model philosophies are presented. 3. The chosen one is then detailed and the corresponding temperature ranges are defined. Tests are thus essential for the verification and to ensure the satellite achieves all the requirements. 7. Consequently. Thanks to the Detailed Thermal Model. thermal modeling involves many assumptions and uncertainties and the design relies on a worst case approach. This behavior can thus only be predicted by thermal analytical methods and can only be fully verify once it is in orbit. tests requirements and methods. Engineering or Development Tests are conducted throughout a satellite design in each subsystem to validate new design concepts or perform measurements to reduce uncertainties. three test categories are established: 1. especially for military space vehicles. the typical test hardware is explained along with the foreseen set-up. unlike electronic equipments that can be ground tested. 2. Acceptance tesing is less severe than qualification testing and 101 .Testing 7 In the previous chapters. many documents have been written to standardize definitions. Those documents also introduced test categorization and levels. proper measures have been taken to ensure those components remain within their temperature limits. Next. the thermal behavior of a satellite is very difficult to verify on Earth.1 Test and model philosophy For several decades. manufacturing errors or any latent defect that would be detected by by normal inspection techniques. the different models revealed that the temperatures of some components go beyond their limits. Today. Flight Acceptance (FA) Tests are conducted to verify conformance to specification requirements and provides quality-control assurance to detect workmanship deficiencies. Qualification Tests are conducted to demonstrate that the design implementation and manufacturing process have resulted in hardware and software that meets specification requirements with sufficient margins. This chapter intends to describe the tests that shall be performed at the Centre Spatial de Liège (CSL). However. First.

However. tight schedules. The above strategies concerned both thermal and vibration tests. a Thermal Cycling Test (TC). a second Pumpkin’s CubeSat Kit have been bought for the engineering tests and model. They can be categorized in three main classifications: Lionel Jacques 102 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . For this reason. thermal test are divided into three tests: 1. 3. Protoflight eliminates the redundancy of building a qualification hardware.Chapter 7. that can be performed under ambient pressure. a Thermal Vacuum Test (TV) is performed under vacuum and subjects the satellite (or equipment) to worst hot/cold temperatures including adequate margins. These alternative strategies introduce higher risks compared to the nominal test program constituted of the acceptance test following the qualification one but increasing safety factors and development tests can mitigate the induced risks. the thermal model relies on several assumptions and uncertainties.2 Tests specifications Now that the tests have been introduced and before describing them in details. 2. enables significant cost saving and is thus well suited for student projects. Protoflight testing accomplishes thus in one test the combined purposes of design qualification and flight acceptance and results from a combination of the two strategies. temperature ranges and margins terminology is presented. It is important to note that items which have undergone qualification testing are usually not eligible for flight because the remaining life regarding fatigue and wear is not demonstrated. Its purpose is the performance verification through functional testing because it is the most realistic ground simulation of the in-orbit environment. Testing are conducted under environmental conditions no more severe than those expected during the mission. However. The main alternative strategy is protoqualification (or protoflight) testing. This is the adopted strategy for OUFTI-1. It has two primary purposes: demonstrate the ability of the thermal control system to maintain temperatures within the specified operational limits and provide data for the TMM correlation. a Thermal Balance Test (TB) is generally conducted as a part of the thermal vacuum test. These three tests can be combined in one test called thermal vacuum cycling test during which the TB test is also performed. alternative test strategies and tradeoffs have been developed to reduce the risks involved by this situation. More specifically. subjects the satellite (or equipment) to a series of cycles of hot and cold temperature plateaus. As previously introduced. 7. The tested protoflight model (PFM) is thus considered eligible for flight. budgetary constraints are such as a dedicated nonflight qualification model is not always feasible. Its main purpose is to reveal latent workmanship defects due to environmental stress. It is therefore on the second CubeSat Kit that measurements of chapter 5 were performed.

The qualification range accounts for a 10 C to 15 C additional margin. . Earth IR and direct solar fluxes depending for instance on solar activity. . Now that the qualification temperature range is clearly defined. clouds coverage. Mathematical models (both GMM and TMM) uncertainties: the temperatures of the spacecraft are governed by a set of nonlinear differential equations for which no closed integration solution is available and numerical solutions are employed based on finite difference methods. safety margins are applied to the worst case predicted temperatures. .Chapter 7. interface conductances. Lionel Jacques 103 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .2 illustrates the different temperature range and margin terminology for European and NASA/JPL programs respectively. They are quite identical in the fact that acceptance range is obtained by adding thermal control uncertainty margins (thermal design margins) and residual margins (FA thermal reliability margin in NASA/JPL). To account for these uncertainties. . Spacecraft physical uncertainties: thermo-optical and thermal properties of the materials. These methods have intrinsic uncertainties: . at which it can be switched on. two other temperatures are used in the following figures and need to be defined: the non-operating temperature (max and min) which are the highest/lowest temperature for an equipment to survive unpowered and the start-up temperature which is highest/lowest temperature of the equipment. Environmental uncertainties: albedo.Conductive and radiative exchange between these isothermal elements are only approximated (Monte-Carlo Ray Tracing. thermal resistance across different section. 3.1 and 7. Qualification temperature range are usually the same as the protoqualification one but additional qualification margins can be added to increase environmental conditions over that expected during the lifetime. . Figures 7. .The definition of finite. To this acceptance temperature. . . ). extra margin are added for protoqualification testing. Testing 1. isothermal elements is physically incorrect. Qualifications margins may also include the cycles duration as well as any other increase in severity to demonstrate the robustness of the design.Numerical integration of the differential equations system is subjected to tolerances and consequently has only a finite accuracy. 2. The resulting temperature forms the basis for the acceptance temperature range.

in conjunction with a simple passive thermal control subsystem.2. Further complications arise when a programme uses certain equipment from a previous project that cannot exactly complement 7. with European programmes usually having the more restrictive temperature limits. a trade-off is made early in each project. either to a. 3. flight acceptance test temp.) Residual margin Environmental design margin 10 ºC Minimum qualification test temperature (1) (2) (3) Thermal design range Thermal control uncertainty 10 ºC or 15 ºC (1) Equipment design environment (2) (3) Notes Figure 7. Maximum qualification test temperature Environmental design margin 10 ºC Upper temperature limit (max. Figure b. allow large temperature limit bandwidths.3 Influence of equipment temperature limits on thermal design The general definition of temperature limits and margins of a spacecraft thermal Thermal design margin design are summarized in Figure B--1. Residual thermal design margin can be zero.1 – limits pertaining to the other equipment. with the resulting costs and effort to qualify the components.3 Design & analysis Testing Thermal design margin ECSS Protoflight/ qualification temperature range B. 4 10 ºC margin does not include test condition tolerances standard [17] Figure B.2.) Residual margin Maximum expected temperature Upper predicted temperature (worst case) Predicted temperature range during service life Lower predicted temperature (worst case) Thermal control uncertainty 10 ºC or 15 ºC Minimum expected temperature Lower temperature limit (min. and the more likely the requirement for an active or semi-active control. flight acceptance temp. the greater the analytical effort is performed to ensure the limits. 15 ºC uncertainty for thermal control unverified. the Thermal margin terminology for NASA/JPL programs [24] Ideally. hot/cold flight acceptance predicted USAF temperature temperature temperature MIL--STD--1540 B range range range Institute of Environmental Sciences Environmental Stress Screening Guidelines Military Standard: Test Requirements for Space Vehicles. The bandwidth between the upper and lower temperature limits defines the thermal design range and the type of thermal FA thermal reliability margin (5°C) design to be used. B. or to impose a narrow temperature limit bandwidth and transfer the costs and the effort to the thermal control subsystem to provide an active control system to maintain equipment within limits. Testing Thermal reliability margin (20°C) FA thermal reliability margin (5°C) ECSS--E--10--03A 15 February 2002 B. The smaller the bandwidth.Temperature limits and margins definitions -1: Lionel Jacques 104 University of Liège 2nd Master in Aerospace Engineering Applied Sciences Faculty 138 Academic Year 2008 2009 .4 Worst case Allowable Flight ESSEH 1981.2 – Thermal margin terminology from ECSS-E-10-03A b Thermal control uncertainty depends on type of equipment: 10 ºC uncertainty for thermal control verified by test. reliability margin (15°C) Thermal The approach in defining limits in certain cases is very different between US and European programmes.Chapter 7.

The spacecraft is then cycled between TQ. The temperature is then increased to the minimum start-up temperature and the equipment is switched on. at the pressure of of 10 4 hPa.min until the number of cycles specified in Table 7.3 is achieved.max ) to accelerate 74 outgassing (baking sequence). When stabilized at the low operating level (TQmin ). February 2002 Testing ECSS Temperature START-UP Mode 2 Mode 1 tE Mode 3 START-UP Mode 1 Mode 3 T (ºC) TNO-max TSU-high TQ-max TAMBIENT tE tE TAMBIENT TQ-min TSU-low TNO-min tE tE tE time (hours) 1 Cycle n Cycles Pressure P (hPa) AMBIENT 1 × 10-4 1 × 10-5 COMBINED CYCLING AND VACUUM TEST time (hours) For explanation of symbols see Table 14 Figure 7. and after the time t E . the test always starts with a temperature increase up to the high non-operating level (TNO.min after which the temperature is raised to ambient conditions and the final functional and performance test can be performed. Then. Nevertheless 6 cycles could be performed.1 As defined above.max and TQ.3 – Thermal vacuum cycling test sequence [17] Figure 7. Functionally tests are again performed during the last cycle at TQ. Lionel Jacques 105 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . After a dwell time t E . All the symbols are summurized in Table 7.max ). the functional tests in cold case are performed.3 Figure 10: Equipment thermal vacuum cycling test sequence sequence.high ) and is then stabilized at the high operating temperature (TQ.max and TQ. the temperature is decreased to the maximum start-up level (TSU.ECSS--E--10--03A 15 Chapter 7. illustrates a typical thermal vacuum cycling test The test begins and ends with a functional test at ambient temperature.min ) during a time t E . Once the temperature is stabilized (time t E ) the functional and performance tests (including TB) are performed and after that it is switched off again to decrease and maintained at the minimum non-operating temperature (TNO. the protoflight approach combines acceptance and qualification: levels are the qualification ones while duration are as acceptance ones [17].

7. For equipment space vehicle. higher gradients Thermal cycling test temperatures are equipment 5. conditions at the low When stabilizedduring orbit. e. At the end of last cycle. some modifications will be provided to meet the particular CubeSat thermal design.16. Normally applicable to the non-operating and maintained at the minimum non-operating temperature (TNO--min) durcondition.3) [17] Table 16: Thermal cycling test parameters (qualification) Parameter Start cycle n (number of cycles) tE (dwell time at Thot/Tcold) a. tE During Dwell last cycle. The temperature rate of change < 20 ºC/min shall apply only to equipment Table 7. b.1. j. h. as shown in Figure 9. For equipment outside the space vehicle.15. Conditions as detailed in applicable design specifications.3 – Qualification and acceptance test levels and durations according to ECSS-E-10-03A standard [17] All the previous curves and tables comes from standards considering conventional spacecraft. Partially functioning. c. Chapter 7.3) [17] 72 Levels Qualification Acceptance margins margins 10 C 5 C Duration Qualification Acceptance 8 cycles 4 cycles Table 7.5 T Symbol Description Test item temperature ECSS Thermal cycling test cycles and duration The equipment shall be tested in the thermal cycling test sequence performing TAMBIENT Ambient temperature 8 thermal cycles.1. Regarding OUFTI-1. TNO-max TNO-min NOTE Characteristic parameters of the thermal cycling test are survive not powered) given in Table 16. k. Lionel Jacques 106 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . b dT/dt (temperature rate of change) Stabilization criterion a Condition / Value Hot 8 2 h < 20 ºC/min 1 ºC/h Thot = TQ-max or TNO-max. After the time tE. the temperature design be decreased to the maximum demonstrates full (hot) start-up level (Tdesign ability) then the temperature shall be stabilized at SU--high) and the TQ-min high operating temperature (TQ--max).2 – Thermal cycling test parameters (qualification) (Fig. 66 b Table 7. Table 14: Legend and symbols for Figures 4. i.16. ture MODE 3 and the equipment item fully energized and fully stimulated). the functional and performance test shall be performed.ECSS--E--10--03A 15 February 2002 NOTE Characteristic parameters of thermal vacuum test are specified in table 15. T The equipmentstart-up be switched off and the temperature increasedwhichto the Minimum shall temperature (lowest design temperature of the equipment.1 within the and symbols for thermal vacuum (and cycling) test sequences (Fig. Testing specified in the appropriate outside thespecifications. g. MODE 1 MODE 2 demonstrates full design ability) Pressure The equipment shall be switched off and the temperature shall be decreased Functionally inert (test item not energized). j. at which the ambient temperature. – Legend space vehicle. TQ-max a Maximum qualification temperature (highest shall temperature at which the equipment After dwell time tE.1. the temperature shall be raised to ambient conditions Switch-off and the final functional and performance test performed. but normally P f. the equipment shall be functionally tested at TQ--max the time Switch-on (Start-up) and TQ--min. at up the SU-low high non-operatingbe switched on) equipment can level (TNO--max). operating level (TQ--min). the Initial and final “functional and performance test” functional test shall be performed. The first thermal cycle begins after the equipment is functionally tested at TSU-high Maximum start-up temperature (highest design temperature of the equipment. The temperature shall be increased to the minimum (cold) start-up temperaapplicable to conditions during launch. Normally applicable to Fully functioning (test switched on. Maximum non-operating temperature (highest design temperature for the equipment to Minimum non-operating temperature (lowest design temperature for the equipment to survive not powered) equipment can be switched on) a. The temperature rate of change < 20 ºC/min shall apply only to equipment within the space vehicle. higher gradients are specified in the appropriate equipment specifications. and after the time tE. Tcold = TQ-min or TNO-min.4 The qualification temperature limits shall be in accordance with subclause 5. The equipment shall befunctional between TQ--max and TQ--min until the number Intermediate reduced cycled and performance test of cycles specified in Table 16 is achieved. ing a time tE. 7. 9 and 10 5.5. design temperature at which the equipment Minimum qualification temperature (the lowest d.

4 – OUFTI-1’s test temperatures definition Table 7. The number of cycles should thus take into account the level of charge of the batteries. Lionel Jacques 107 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . The predicted temperature of the structure ranges from -30 C to +35 C but measures have been taken to ensure that the temperature of the batteries does go beyond their limits.Chapter 7.3 Two solution are proposed to deal with this problem: 1. identical to the discharge one is only -20 C to +60 C while their operating one is reduced to 0 C to +45 C. Predicted (op) Min Max 5 C 35 C -30 C 35 C Qualification Min Max 0 C 45 C -40 C 45 C Non-operating Min Max -20 C 60 C -45 C 85 C Start-up Min Max 5 C 35 C - Batteries Other Table 7. Then raise the temperature to the low startup temperature TSU. 2. Testing OUFTI-1’s TMM results and operating/non operating (NO) temperature ranges showed that the batteries were the more critical components. Here again.low before switch on the CubeSat and decrease its temperature again to the operating one. This is in contradiction with the above figure 7. On another hand.min instead of the non-operating one. the batteries should now range between +5 C and +35 C. The second solution consist in decreasing the temperature down to the non-operating one of the batteries during the first cycle with the CubeSat switched off. all the other units (structural and electronic) have a wider operating (and non-operating because they are identical) range at least equal to the -40 C to +85 C industrial temperature range.4 confirms that the batteries are well the more restrictive components. the heater are designed to maintain the batteries above 0 C when the temperature of the structure is -30 C. the CubeSat should be kept switched on to allow the heating system to maintain the batteries above 5 C and the number of cycles should be defined by taking into account the level of discharge of the batteries resulting from the heaters consumption. The dissipation transistor whose temperature range is even wider: from -65 to +200 C (operating or non-operating). A separate test should be performed on the batteries for their nonoperating qualification. The CubeSat should also be operational (at least the heaters) during all the subsequent cycles. Their NO temperature range. The first solution is therefore to keep the CubeSat switched on after the first functional test and decreasing the temperature down to the low qualification operating temperature TQ. Thanks to the active and passive thermal control system. The non-operating lower temperature of the batteries (-20 C) is thus higher than the operating temperature of the CubeSat (-30 C): the temperature of the switched off CubeSat could only decrease to -20 C while if it is switched on.

. vibrations testing which are not discussed in the present thesis (but in the STRU subsystem [42]) are also required and usually performed before thermal testing. 2m. some answers can already be given. Development tests demonstrate that such pressure levels were harmful to the batteries that inflates under mechanical stresses. Actually. 3m and 5m).5 Lionel Jacques 108 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . The key parameter is to ensure that the mean free path of remaining molecules in the chamber is greater than the chamber characteristic size (diameter). The following questions need to be answered: How is fastened the CubeSat in the vacuum chamber? How to impose the temperature variation on the CubeSat? How the low pressure levels are achieved and is-it harmful to the CubeSat? Is solar simulation feasible? What are the functional test that will be performed and how to collect the data? Shall the batteries be charged during the tests? . the lower the required pressure. Nevertheless different heating and cooling methods are available each with its own advantages and drawbacks. CSL has six vacuum chambers called FOCAL (Facility of Optical Calibration at Liège) whose diameter range from 0.3 Test set-up Thermal testing involves specific equipments such as vacuum chambers allowing active thermal control. The pumping phase is divided in two stages: a first one performed through classical pump down to 10 2 mbar and the second using a turbomolecular pump to decrease the pressure down to 10 5 mbar. Both will be performed at the Centre Spatial de Liège (CSL) which is one of the four Coordinated European Environmental Test Facilities (CETeF) of the European Space Agency (ESA). the required pressure level is 10 5 hPa. the pressure required depends on the size of the vacuum chamber: the bigger the chamber. While some of these questions will remain partially unanswered because the overall design is not yet defined. There is no universal method. As explained in test specification section. The first two questions are not independent because the way a satellite or instrument has its temperature imposed influence the way it is fastened to the chamber and conversely. Each thermal tests set-up has its own dedicated set-up and heating/cooling method. Moreover. many test parameters have still to be defined in addition to the different temperature ranges of the thermal vacuum cycling.5m.25 to 6.. While the satellite Planck have been tested in FOCAL 5 during nearly two months. the foreseen vacuum chamber for OUFTI-1 thermal testing is FOCAL 1.5m (with 1. Testing 7. Nevertheless.Chapter 7.5. They can be classified in two categories: radiative and conductive methods and are presented in Table 7.

Accurately simulates solar environment . However. Sometimes. The CubeSat has five of its six faces covered with solar cells. The internal sides of the shroud are panted in black (here it is black opened honeycomb) to enhance exchanges between the satellite and the shroud. Testing Radiative Method Solar simulation Heating lamps Heater plates Conductive heater Advantages .4 shows the shroud used for Planck testing in FOCAL 5. The heater plates assembly is covered with MLI to reduce exchanges and losses with the walls of the chamber remaining at ambient temperature. because of its cubic shape. providing good flexibility .Parallel illumination required large set-up . booms etc.requires knowledge of absorbed fluxes and cooling in heater plates .5 – Thermal testing heating/cooling methods [24] Solar simulation would have been appropriate to simulate the real thermal loads on the CubeSat and would have allowed the use of the solar cells. it seems convenient to use of heater plates. It is not available for the moment at CSL.Chapter 7.Allow GMM error detection .direct and independent surface control Table 7. seeing the relative low thermal inertia of the CubeSat. such measure should not be necessary. However. Moreover. ConseLionel Jacques 109 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .environment accurately known and allow independent surface control .Cannot simulate nonsolar heat loads . The plates and MLI constitute the so-called thermal shroud.many lamps required and interfering with each other . Figure 7. However solar simulation is difficult to implement because the heat generated by the halogen is difficult to cope with unless the lamps are relatively far from the other cooler environment elements.lamps can be placed efficiently and are independent. the continuous rotation of the CubeSat involves relative uniform external temperatures of the sides and we can assume that all the faces have the same temperature. All the heater plates surrounding the CubeSat can then be controlled by a unique fluid loop.requires knowledge of absorbed fluxes and test blanket are required of heaters are mounted to them .Does not assume a prior known environment . The fluid loop consist of a nitrogen thermally controlled inlet line that impose the temperature of the plates. thermal straps between the satellite and the shroud are used to enhance heat exchange and increase the heating or cooling phase.limited flexibility heater plates .good for appendages such as antennas. one for each face of the CubeSat. Drawbacks Solar simulation chamber not available at CSL .

Chapter 7. monitoring and conLionel Jacques 110 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . the mounting interface should be insulating from the chamber walls ambient temperature. all the parameters must be available to allow a comprehensive understanding. the only remaining face available to fasten the CubeSat in the chamber is the antennas’ face. At any time.5 – Thermal vacuum cycling test set-up Figure 7.5 also shows that a way to communicate with the CubeSat and collect data must be provided.4 – Example of thermal shroud used for the testing of Planck in FOCAL 5 quently. Testing Figure 7.5 illustrates the foreseen set-up with fluid loop controlling the shroud. Thermal vacuum chamber Temperature controlled shroud CubeSat Data acquisition path and control TC/TM Reference point Mounting interface Temperature controlled fluid loop Figure 7. Again. Figure 7.

we can already assume that the following tests should probably be conducted: Each defined mode should be tested. the amplifier shall be used normally and its temperature shall be carefully monitored especially during the hot case. If necessary. separate tests should be conducted. As the CubeSat is inside the vacuum chamber. especially the temperatures to allow the correlation of the Thermal Model. these sensors will be used as reference point to monitor the batteries temperature. The dissipation system shall be switched on with full power. Nevertheless. communication through the antennas are useless and suitable testing procedures shall therefore be defined without the use of the antennas.Chapter 7. However. Among these. during transmission. A temperature sensor shall be located at the reference point. In any case the temperature of the batteries would go beyond their limits. However. the shroud temperature are controlled by the operator in such a way that the reference point temperature follows the thermal cycling specification curve. full D-STAR transmitting and receiving. 2. two reference points are defined and located as follows: 1. During the test. The thermal cycling curves presented above require to define a reference point temperature that is representative of the mean temperature level of the equipment. Therefore. on the structure. In the cold case. they are difficult to be clearly defined at this development phase. The sensor must obviously not be located on solar cells nor anodized rails. Testing trol of the CubeSat behavior. the more convenient place is again the antennas’ face (face 4). This reference point shall be representative of the global temperature of the CubeSat. the heating system of the batteries shall demonstrate that the design implementation fully meets the requirements. Concerning the performance and functional tests that are performed during the first (TB) and last cycles. additional power is provided through the USB interface to meet the predicted dissipated power with 10% margins [? ]. As the batteries will already be equipped with temperature sensors. the thermal vacuum cycling test will be performed in stowed configuration. The dissipation transistor and batteries temperatures shall carefully be monitored. the second reference point is the batteries. In the present case. it shall be placed far enough from the mounting interface. As a comprehensive test is not possible. proper measures shall be taken to bring it back within its safe range. All the housekeeping parameters shall be monitored and saved. under ambient pressure: Lionel Jacques 111 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Moreover. Extensive beacon functional and performance tests shall be conducted in hot and cold cases.

7. a detailed test description was tot possible. Once the CubeSat will be fully designed.4 Summary Thermal testing is a compulsory step in the development of a satellite. For this reasons. Concerning OUFTI-1. As the CubeSat development is not terminated.Chapter 7. However. Testing Simulations have shown that the antennas could undergo relatively high temperatures (up to more than 200 C at their free extremity). foreseen set-up and functional tests have been described. Heaters should be designed to cope with the convective losses. Lionel Jacques 112 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . specific functional tests should be conducted in the deployed configuration with dedicated thin heaters located on the antennas’ free extremity. the tests shall be conducted at the Centre Spatial de Liège. a test procedure document shall be written in agreement with the Centre Spatial de Liège and all the subsystems to carefully define the functional tests that should be performed on the CubeSat.

contact phenomenon in the structure have been confirmed through IR imaging. the DTM showed that insulating the batteries is required. The heaters will be switched on when the temperature reached 5 C. As the thermo-optical properties of the structure and thermal properties of the batteries were subjected to some relatively high uncertainties.Conclusions Throughout this work. involving a 10 C temperature decrease of the batteries (from 48 C to 38 C) and 20 C of the transistor (from 115 C to 95 C) . as the battery model is not yet fully defined. simulations showed that the batteries would undergo too low temperatures and that they were out of their temperature ranges when coming out of the eclipse. The first models brought to light that issues would occur in both cold and hot cases. the heaters have not yet been ordered. On another hand. Concerning the hot case. drastic measures had to be taken in order to reduce temperatures: we showed that a thermal strap encircling the dissipation transistor would be required and fastened to antennas’ deployment mechanism panel. Nevertheless. A heating system has therefore been designed: two 250 mW Minco heaters. In the cold case. will ensure that the temperature of the batteries will remain within its allowable range. depending on the dimensions of the batteries. the dissipation system developed by EPS subsystem was the cause of nearly all issues: involving too high temperatures on the batteries and the dissipation transistor. measurements have also been conducted through which the heat capacity and transverse conductivity of a typical LiPo battery were determined. only the Detailed Thermal Model enabled a comprehensive investigation of these issues. as well as the emissivities of the two coating of the aluminum frame. different thermal models with different purposes and complexity have been carried out. set in parallel combined with three temperature sensors (to avoid the problem encountered by Compass-1). It has been presented that both hot and cold cases had a strong impact on the design of the CubeSat. To reduce losses and increase its efficiency. relocating 113 . Even in the worst case. we contact the company Moss Plastic Parts to order some sample nylon washers because creep phenomenon should be carefully studied to ensure structural integrity of the batteries’ PCB fastening. Moreover. To solve these problems. we verified that the consumption of the heating system remains acceptable compared to the capacity of the batteries: up to 150 mAh compared to the 100 mAh (and therefore 2000 mAh for both batteries) typical capacity of one LiPo battery. In addition to that. However. Covering the batteries with a low emissive tape such as aluminum thermal tape is planned and Nylon washers combined to titanium spacers shall be used to fasten batteries’ PCB. Those results were therefore included in the DTM.

At last. Bob Twiggs. The Minco resistances have been ordered to evaluate the feasibility of their integration on antennas’ panel. Finally. working within a team is the main one and has been extremely constructive. the CubeSat leitmotiv should be: Test. EADS Astrium.75 W assumed for classical amplifiers. because of its quite low efficiency and therefore high heat dissipation. which has been confirmed. Working within a CubeSat project has numerous advantages. However. this project allowed us to present the project in different international events and workshops during which we met other CubeSat teams and exchanged ideas. we gained an invaluable hands-on experience in satellite design. Azur Space and many others. especially in thermal design. Among them. using the ADL5541 could reduce the temperature from 70 C to 45 C because the ADL5541 only dissipates 0. Particularly. we presented the project on the Space Days event in Liège and at the Redu Eurospace Center. Tests should also be conducted on the heat rejection design to verify that the strap and relocated resistances play their role.Conclusions a part of the dissipated power through two Minco resistances (reference XHK5377R26. The third issues occurring in the hot case involves the COM amplifier. its efficiency should be taken into account when selecting the amplifier. the heating system should be extensively tested. in our opinion. No concrete measures have been taken to solve this problem because too many uncertainties still surround its definition. Thales Alenia Space. we stated that the amplifier should be localized either close to one corner of the COM PCB or it the center of the PCB. Moreover. it allows a global view of the project and highlights the strong interaction that may occur between the different subsystems. batteries’ PCB should be fastened to EPS2 PCB instead of EPS one because it contains less dissipative components and is therefore cooler. we participated to the Second European CubeSat Workshop at ESTEC. Namely. CubeSat project also allows to work within an industrial framework and to collaborate with famous companies or agencies such as the European Space Agency (ESA). for reliability purposes.5 W instead of 1. locating the amplifier on one edge should be avoided. The effect of hot antennas should be investigated to see the effect on their radiation pattern because the vacuum tests that will be performed et the Centre Spatial de Liège will probably be in the stowed configuration. numerous tests should be conducted to correlate the model and because this is the only way to ensure that the design meets the specification requirements. For instance. according to Prof. respectively reduced to 35 C and 67 C. test and test again ! Lionel Jacques 114 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Concerning the future developments. Furthermore.3L12B) in parallel also fastened to the antennas’ panel involves another 3 C and 30 C temperature decrease of the batteries and transistor. working on this project has been very enriching and its primary goal has been achieved: through it. On the educational point of view.

Appendix 115 .

Appendix A Acronyms ADCS A&M BOL COTS CSL D-STAR DTM EOL EECS EPS ESA ESATAN ESTEC FE FEM FHTS FOCAL GMM GND IC MCRT MLI OBC OUFTI P-POD PCB REF SSETI STM STRU TCS TMM ULg Attitude Determination and Control System Department of Aerospace and Mechanical Engineering Beginning Of Life Commercial-Off-The-Shelf Centre Spatial de Liège Digital Smart Technologies for Amateur Radio Detailed Thermal Model End Of Life Department of Electrical Engineering and Computer Science Electrical Power Supply European Space Agency European Space Agency Thermal Analysis Network European Space Technology Research Center Finite Element Finite Element Model Fluid Heat Transport System Facility for Optical Calibration at Liege Geometric Mathematical Model Ground Integrated Circuit Monte Carlo ray-tracing Multi-layer Insulation On-Board Computer Orbiting Utility For Telecommunication Innovation Poly-Picosatellite Orbital Deployer Printed Circuit Board Radiative Exchange Factor Student Space Exploration and Technology Initiative Simplified Thermal Model Structure & Configuration Thermal Control System Thermal Mathematical Model University of Liège OBC PCB OBC2 PCB EPS PCB BAT PCB EPS2 PCB COM PCB main on-board computer printed circuit board secondary on-board computer printed circuit board main electrical power supply printed circuit board secondary printed circuit board on which the batteries are attached innovative electrical power supply printed circuit board communication subsytem on-board computer printed circuit board Lionel Jacques 116 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

5 = = = 12.04094 Then.Appendix B Aluminium frame faces links This appendix explains how the links between the different nodes of the frame have been computed. left) and R1 is equal to one half of the difference of temperatures divided by the resulting flux: R1 = 1 ∆T 0.21 [K/W ] GL 2Qtot 0. R1 along the diagonal and R2 along the horizontal/vertical. a SamcefField finite element analyses gives the resulting flux from which the thermal conductance can be deduced. The thermal resistance/conductance does only depends on the geometry and conductivity of the frame links. the diagonal does not get involved in the conduction path (figure 7. The equivalent resistance is therefore equal to: Lionel Jacques 117 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 . right). 𝑅′1 𝑅𝑐 𝑅1 𝑅′1 𝑅2 𝑅1 𝑅𝑐 Figure 6 – Conductance model of face 3 R1 and R2 are easily determined: the first one is obtained by considering a quarter of the face and setting the boundary conditions on both extremities. The method is the same as the one used for the computation of the global links in the high level model : by imposing the temperature gradient across the desired geometry. Face 3 Figure 6 describes the resistances network (in red) representing the heat flow path between the nodes of the face. R2 is obtained by involving the diagonal in the flow path and setting one of the two temperature at its extremity (figure 7. By this way. For symmetry reasons. there are only two different resistances.

14 = 23. This means that instead of two different resistances. there are three ones: R1 .81 [K/W ]. the boundary conditions were set around the screw holes. To take into account constriction effect around the screw hole. after some simplifications. between to opposite screw holes.21 = 2.62 [K/W ] and R3 = 37. One finally gets R1 = 24. the heat flow passing from face 3 to face 1. one can perform three FE simulations to obtain a set of three equations for the three unknowns: one for R1 + R2 .81 + 21. Using the superposition principle.95 [K/W ] while the finite elements global resistance. R2 = 24. the second for R1 + R3 and the third one for R2 + R3 . is 23. R2 and R3 as shown on figure 8.04 [K/W ]. This leads to Rc can then be approximated by differentiating R1 and R1 0 RC = R1 R1 = 15. Hence. Lionel Jacques 118 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .29 [K/W ] with Req = 21. 2 and 5 Face 1.14 [K/W ] Figure 7 – Face 3 conduction SamcefField results The previous model assumed perfect contact on all the bracket surface but experiments of chapter 6 showed the presence of contact phenomenon. To determine this constriction resistance. an additional resistance Rc (in blue) is considered between the nearby node and the screw hole.55 [K/W ]. 2 and 5 are identical and but no longer symmetrical.76 [K/W ] Faces 1. representing the bracket fastened to face 4.5 or 4 shrinks around the contact area. Finally.02 12.2. from one side to the opposite one is equal to: Rtot = Rc + R1 + R1 //R2 = 2. 0 The computed flux correspond to the green resistances R1 of figure 6. close to the screw.Appendix Req = R1 + R1 //R2 = R1 + R1 ( R1 Req ) R1 R2 ) R2 = R1 + R2 Req 2R1 One finally gets: R2 = 33. 0 . the total transverse resistance. Face 6 Face 6 conductances are identical to face 3 except for the link with the additional node 3610.

3400 series 5.06 8.19 7-6 3.06 7-6 2.00 8.69 7-9 64.Appendix 𝑅1 𝑅3 𝑅2 Figure 8 – Conductance model of faces 1. All the conductance are summarized in Table 6.69 3-9 2.00 8.00 7-9 8. 2 & 5 .06 3-4 4.00 5-9 3.19 7-8 62.19 7-8 Face 4 . nodes 1-2 1-9 1-8 1-2 1-9 1-8 1-2 1-9 1-8 1-2 1-9 1-8 2-10 GL nodes GL nodes GL nodes Face 1.3300 series 8.19 3.69 7-9 4.00 3-9 3.19 3.19 8. 2 and 5 Face 4 Face 4 links have all been determined by the same way.07 7-8 Face 3 .7 GL 4. 3200 & 3500 series 4.19 5-4 8.19 7-6 3.3600 series 8.8 5-4 4.00 7-9 8.69 5-9 2.06 5-6 4. in [W/K ] Lionel Jacques 119 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 .07 3-2 4.3100.19 3-2 8.56 5-9 2.00 5-9 3.19 3-4 8.00 3-9 3.19 3-4 8.9 3-4 3.19 5-6 8.69 4.17 3-9 5.50 3-2 10.19 3-2 8.07 7-8 Face 6 .00 8. through FE models.19 3.85 5-6 4.06 7-6 3.19 5-4 8.19 10−2 Table 6 – Aluminium frame faces conductances.19 5-6 8.07 2.07 5-4 4.19 8.

z ) where the function G ( x. Figure ?? illustrates the different notation used. z) = with y a+b c 2π a = (y b = (x c= η) ξ) (x (y ξ )2 + z2 arctan η )2 + z2 arctan ξ )2 + ( y η )2 + z2 η ξ )2 + z2 (x x ξ η )2 + z2 (y z2 log ( x 2 The view factors with the lateral aluminum panels representing the lateral frame were indirectly deduced from the reciprocity properties of the view factors (Ai Fij = A j Fji ) and from the energy conservation (∑ Fij = 1 for a closed system) that is enforced by setting the aluminum panels such as they form a closed environment with the PCBs. y. Lionel Jacques 120 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . y j . η. The view factors between each elements of the two opposite PCBs were indeed required in the Gebhart’s formula. ξ l . 𝜂 𝜂1 𝜂2 𝐴2 𝜉1 𝜉2 𝑧 𝑦1 𝐴1 𝜉 𝑦2 𝑦 𝑥1 𝑥1 𝑥 Figure 9 – View factor between two parallel rectangles [15] F= ( x2 1 x1 )(y2 ∑ y1 ) i∑ j∑ k=1 l∑ =1 =1 =1 2 2 2 2 ( 1) i + j + k + l G ( x i . ηk .Appendix C Analytical computation of VF between parallel rectangles Here is the formula used for the view factor computation in the verification process of DTM’s Geometrical Mathematical Model. ξ.

the global one: on the left is the heat fluxes computation module already presented and on the left the thermal lumped parameter model which is described in the second diagram. Within this. First. HF 1 DCM r_eci r_sun DCM HF 1 HF 2 T_OBC 2 u+b °K to °C 1 T_OBC2 r_sat _ECI HF 2 HF 3 T_OBC r_sun _ECI HF 3 t JD sun Cs Cs HF 4 HF 4 u+b °K to °C 1 2 T _OBC 0. The results obtained with this model are relatively well correlated with ESATAN/ESARAD ones. the PCB stack block is detailed in the third and last module.Appendix D Advanced Simulink Thermal Model Here are the advanced Simulink model diagrams.3 255 albedo _coeff HF 5 HF 5 T_Bat HF 6 T_Earth HF 6 u+b °K to °C 2 Incident Heat fluxes 3 T_Bat P_COM 0 P_EPS2 T_EPS u+b °K to °C 3 4 T _EPS 0 P_EPS P_ELEC P_OBC 2 T_EPS2 u+b °K to °C 4 5 T_EPS 2 0 P_OBC 0 T _Face 1 0 T _Face 2 P_Bat T_COM u+b °K to °C 5 6 T _COM 0 T _Face 3 0 T _Face 4 0 T _Face 5 u+b °K to °C 6 9 T_Faces Thermal Model Electrical _Power _Face 1 Electrical _Power _Face 2 Electrical _Power _Face 3 Total _Power _Face 1 Total _Power _Face 2 Total _Power _Face 3 Total _Power _Face 4 7 P_ELEC P_ELEC Total _Power _Face 5 Total _Power _Face 6 8 P_TOT 0 T _Face 6 Electrical _Power _Face 5 0 Electrical _Power _Face 6 Figure 10 – Advanced Simulink thermal model main high level diagram Lionel Jacques 121 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

Appendix GL 2/4 B A 3 HF3 Heat Fluxes 3 Deep Space Temperature Deep_Space _Temp Face 3 3 f(x)=0 Solver Configuration GL 3/4 B A Temp Face A B GL 3/2 B GL 3/6 A Face 2 HF2 4 HF4 Heat Fluxes Temp Face 4 7 P_COM P_COM T_BAT 3 T_Bat B T Temp Face 2 Heat Fluxes 2 Face 4 8 P_EPS2 P_EPS 2 T_COM A PS S 6 T_COM GL 4/5 A GL 3/5 B GL 6/4 A 9 P_EPS B A B B A GL 6/2 A B GL 3/1 A B P_EPS T_EPS 2 A B T PS S 5 T_EPS2 GL 2/1 10 P_OBC2 P_OBC 2 T_EPS A B T PS S 4 T_EPS 11 P_OBC P_OBC T_OBC 2 A B T PS S B T 1 T_OBC2 Face 1 12 P_Bat 5 HF5 Heat Fluxes Temp Face 5 P_Bat T_OBC A PS S 2 T_OBC Temp Face 1 Heat Fluxes HF1 1 PCBs Stack Face 5 GL 3/3 B A B A Temp Face 6 A B GL 6/5 GL 6/1 Face 6 Heat Fluxes HF6 6 B A GL 1/5 Figure 11 – Advanced Simulink thermal model low level diagram: global model Lionel Jacques 122 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

Appendix COM T_COM 1 P_COM S PS S A 1 B PS S B B B B 2 T_EPS2 A B 3 T_EPS A B B A B A B A A EPS2 A A Battery 1 top B B B A A A A Battery 1 bottom B 2 P_EPS2 S PS S A B T 3 P_EPS S PS S A 0.5 B EPS A 1 T_BAT B B B A A A BAT 4 P_OBC2 S PS S A 4 B OBC2 T_OBC2 A B Battery 2 bottom B B B A A 5 P_OBC S PS S A A B T A A B Battery 2 top B 5 T_OBC OBC PS S 0 S PS 6 P_Bat Switch S A B Figure 12 – Advanced Simulink thermal model low level diagram: PCB stack model Lionel Jacques 123 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .

n + Q4. If there are B temperature boundary conditions and therefore I = n B unknowns. Here. 4 𝐺𝐿𝑦 Δ𝑦 𝐺𝐿𝑥 1 𝐺𝐿𝑦 2 𝑛 𝐺𝐿𝑥 3 Δ𝑥 Figure 13 – Equivalent conductivity of BAT PCB fastening in function of the PA washer thickness and thermal conductivity of the spacer Generalizing for n nodes the equations can be set under the following matrix form AT=g where A is a n n matrix. Figure 13 illustrates the incoming fluxes exchanged between adjacent nodes and energy conservation leads to: Q1. one has: GL x ( T1 Tn ) + GL x ( T3 kt∆ Tn ) + GLy ( T2 Tn ) + GLy ( T4 Tn ) = Qn where GL x = ∆xy and GLy kt∆yx are respectively the thermal conduc∆ tances along x and y directions with k the thermal conductivity and t the thickness of the plate. j) representing the conductive coupling between nodes i and j and g(i ) the external loads of node i.   . A(i. the system can be divided as follows:   T1  .   TI  TI  T=  TI +1  = T B    .  .n + Q3. the case a 2D plate will be considered but the method can easily be generalized to 3D problems.n + Qn = ρc Tn In case of steady state. Concerning thermal conduction. The lumped parameter method relies on a discretization of the domain into smaller isothermal elements called nodes.   .Appendix E 2D conduction lumped parameter method The lumped parameter method allows to transform differential equations into algebraic equation. Tn Lionel Jacques 124 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .   . Tn = 0 and using electrical analogy. For rectangular elements.n + Q2. the algebraic equation are moreover linear (if materials properties do not depend on temperature) and can therefore be written under a matrix form.

this gives: AI I A BI A IB A BB TI TB = g I g B and the unknowns temperature T I are deduced without inversing the entire A matrix: T I = A−1 g − A IB T B II I The method has been correlated with SamcefField has shown in Figure 14. Figure 14 – Equivalent conductivity of BAT PCB fastening in function of the PA washer thickness and thermal conductivity of the spacer Lionel Jacques 125 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 − 2009 .Appendix Splitting matrix A and vector g in the same way.

(Quoted page 19. “From mission analysis to space flight simulation of the OUFTI-1 nanosatellite.” http:// www.arianespace. (Quoted pages viii. vol. Chou. 2009. fh-aachen. (Quoted pages xii. revision 0.UM-ESARAD-024. 5.) [2] ESARAD User Manual .delfic3.) [13] “Delfi-C3’s website. “Azur space solar power gmbh.” Master’s thesis.) [11] J. Contrôle thermique des engins spatiaux. “50 mhz to 6 ghz rf/if gain block. [7] Azur Space Solar Power.Bibliography [1] I. (Quoted page 18. Chiarello.de/. pp. adl 5541.azurspace. Inc. Belgium. Issue 3. mar 2006. France.) [6] Vega User’s Manual. Leicester LE8 6LH United Kingdom. “Design properties for state-ofthe-art thermal control materials for manned space vehicles in Leo. Whetstone. adf 7021. revision 0.) [5] ——. University of Liège. 47 et 75. C.” Acta Astronautica. B. 369–379. Jones. 9. University of Liège. (Quoted page 23.com. (Quoted page 8. (Quoted page 42. [4] Analog Devices. com. 177.” 2007. Beukelaers. Cambridge Road. David. Cambridge Road.” 2007. Jun 2004.) [10] L.” May 2007. http://www. (Quoted page 21. 91006 EvryCourcouronnes cedex. H.nl/. “Design of Thermal Control Sub-System (TCS).) [3] ESATAN User Manual .) [12] “Compass-1 Phase B documentation..P.UM-ESATAN-004. 1998.) [9] V.) TM TM 126 . 23 et 76. Arianespace. Sept. and K. École nationale Supérieure de l’Aéronautique de l’Espace.” http://www. Alet. (Quoted page 42. W. Picosat course.) [8] H.” Master’s thesis. Leicester LE8 6LH United Kingdom. “High performance narrow-band transceiver ic data sheet. (Quoted pages ix et 43. 1996. oct 2007. Babel.” http://www. Alstom Aerospace. Belgium. revision A. Whetstone. Boulevard de l’Europe. (Quoted pages 86 et 99. 39.raumfahrt. 42. 2009. Alstom Aerospace. oct 2007. “Design and implementation of the control and monitoring system for the ground station of nanosatellite OUFTI-1 of University of Liège.

” Master’s thesis.” http://www. G. Jun 2004. “DTU Satellite Systems and Design Course. (Quoted pages xii. 7. 70. “Mission design for the cubesat oufti-1. (Quoted pages x. Henrard.) [21] ThermaCAM Researcher User’s manual. University of Liège. “Summary of Properties for Kapton R Polyimide Films. 2009. Cranfield University. Swinerd.space engineering : Testing.Bibliography [14] DuPont .) [28] R. 16. 23.25 à bord du nanosatellite OUFTI-1. “View Factors for Perpendicular and Parallel.” ESA Publications Division ESTEC. vol. Hardy. (Quoted pages xi et 120. PSS-03-108.) [26] J. 61 et 63.D. 2009. low Earth orbit spacecraft. “Spacecraft thermal control design data. J. Gilmore. [23] S. 2nd ed. 71.” Ph. March 2003. Smith.1. (Quoted page 42.pdf.dupont.” ESA Publications Division ESTEC. Hannay. “Développement de l’infrastructure de mesure du nanosatellite OUFTI-1. (Quoted page 10. School of Engineering. Ehlert and T. “Mission Design for the CubeSat OUFTI-1. 1993. 2008. 42. 7.) [27] P.dsri. University of Liège. Evrard. 105 et 106. R. rev.) [25] S.” Journal of Thermophysics and Heat Transfer 1993. Belgium. xii. Dec 2006. Belgium.com/Kapton/en_US/assets/downloads/ pdf/summaryofprop. 173–174. ix. Fortescue. 8 et 76. (Quoted pages ix.” Master’s thesis. Sep 2001. xii. Belgium.” 1989. dissertation. Gramme Institute. 47 et 48.) [22] P. [17] “ECSS-E-10-03A .) TM TM Lionel Jacques 127 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . Gramme Institute. 3rd ed.) [19] N.” Master’s thesis. (Quoted pages viii.) [20] H. Galli. Flemming.” Master’s thesis. 104 et 109. Belgium. 2006. 2009. FLIR Systems. Harkness. “Implémentation du protocole AX. (Quoted page 42. 104.) [15] J. Stark.space product assurance : Data for selection of space materials and processes.” Master’s thesis. (Quoted pages viii. 2009. (Quoted pages xii et 47. (Quoted pages viii. [24] D. Rectangular Plates. December 2002. x. “An aerostable drag-sail device for the deorbit and disposal of sub-tonne. “Modeling of the Attitude Determination and Control System of the nanosatellite OUFTI-1’.) [16] ESA-ESTEC. (Quoted pages viii et 10. Feb 2002. Cubesat Thermal Design. Spacecraft Thermal Control Handbook: Fundamental Technologies. 42. Spacecraft Systems Engineering. and G.” www2.) [18] “ECSS-Q-70-71A . 32 et 52. Aerospace Press. Wiley. 25.dk/roemer/pub/cubesat. University of Liège. pp.

Tech. (Quoted page 49.” http://www.” www. Belgium. with Amendment 1. Sep 2003. EPFL . May 2009. Inc. 8 et 76. 50.” University of Liège.wbdg. Rep. (Quoted page 60.LTCM..) [40] Minco. (Quoted pages viii. [30] L.org/ ccb/FEDMIL/dtl5541f.SwissCube thermal Analysis.. “Superior Lithium Polymer Battery. “Phase C . (Quoted page 21. “Design and Implementation of On-board Digitaly Controled Electrical Power Supply of Student Nanosatellite OUFTI-1 of University of Liege. (Quoted page 48. Belgium.) [36] F.) [42] G. “Flexible Heaters Design Guide. University of Liège. specification SLPB723870H4.) [39] “Military specification .” http:// swisscube. University of Liège.” Department of Aerospace and Mechanical Engineering. “PC/104-Plus Specification. Rep.” http://www. Tech. 2009. Oct 2008. University of Liège. wbdg. “Pic18f2331/2431/4331/4431 sheet.) [37] Microchip Technology. Madhour.ch/.” Master’s thesis.3.) data [38] “Military specification . 77 et 87.” Master’s thesis. Belgium. specification for SLPB554374H..minco. “Design and Implementation of On-board Telecommunication System of Student Nanosatellite OUFTI-1 of University of Liège. Jacques and G. University of Liège.epfl.) [41] PC/104 Embedded Consortium. (Quoted page 21.org/ccb/FEDMIL/a8625.) [35] Y. 2009. “Superior Lithium Polymer Battery. Switzerland. Jun 2008.” Mar 2007. “Rapport de stage : Etude thermique d’un capteur cylindro-parabolique. (Quoted pages 93 et 94. Belgium..” www. Tech. Belgium. Ltd. Pierlot. version 2. University of Liège. 92 et 93. for Aluminum and Aluminum Alloys. nov 2008.” Master’s thesis.. Rep..) [32] ——. [31] Kokam Co. Lausanne. Rep. mar 2009. “OUFTI-1 : Flight system configuration and structural analysis.Centre Spatial de Liège. Ledent.) [34] ——.MIL-DTL-5541F Chemical Conversion coatings on Aluminum and Aluminum alloys.” Apr 2007. Nov 1990. Tech. Tech. 2009. (Quoted pages 8. (Quoted pages x.Bibliography [29] L.pdf. “Spécifications mécaniques des cartes pc104 destinées au projet oufti-1. (Quoted pages x et 87.com. Pierlot... Belgium.pc104.org. Rep.) for [33] P.) Lionel Jacques 128 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 .pdf. 10 et 108. (Quoted pages viii. Jacques. Mahy. Belgium.” 2003. Liège.MIL-A-8625F Anodic Coatings. (Quoted page 60. “Test report .Vacuum tests for Varta and Kokam batteries.” University of Liège .

“2N5151 High Speed Medium Voltage Switches Datasheet. Northfield. Feb 2008. Van Noord.cubesatkit. Poinas. “CubeSat Kit FM430 Flight Module Datasheet. umich. 2009. “VARTA Sales Program and Technical Handbook.com. Fundamentals of Astrodynamics and Applications .) [52] Sheldahl Red Book .TOS-MCT. “Conception d’expériences spatiales. 2005. Inc.” mar 1997. (Quoted page 21. (Quoted page 21. http://www. 42 et 70.) Lionel Jacques 129 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . (Quoted page 16. Rep. pages 16.ESTEC .” Master’s thesis. (Quoted page 69. (Quoted pages 8. Teney. “Mission Design for the CubeSat OUFTI-1. Tech.epfl. Nov 1999. Apr 2006.) [57] J. Microcosm Press and Kluwer Academic Publishers. (Quoted page 23. issue 2.Experiment Thermal Interface Mathematical Models.” ESA . oct 2007. “GLAST Large Area Telescope Calorimeter Subsystem Thermal Design and Analysis. (Quoted page 18. “Pressure Sensor Development Using Hard Anodized Aluminum Diaphragm and Thin Film Strain Gauges. “Satellite Thermal Control Engineering.. “CubeSat Kit User Manual. “Online calculator for the thermal properties in the plate direction for multi-layer glass-epoxy PCBs.Aerospace Thermal Control Materials.” Jun 2008. Belgium. (Quoted page 42.varta-microbattery. reference : RO-EST-TN-3021. Rep. 2009. Ecole polytechnique. Prat.) [45] P. (Quoted pages 6 et 21. University of Liège. 1150 Sheldahl Road.) [54] D. Jun 1999. Belgium.) [44] ——.) (Quoted TM [51] Semelab. 50 et 87.) [50] P. Mar 2003.” http://www.” http://data.) [48] ——. Tech.” Master’s thesis.) [58] VARTA Microbattery GmbH. Vallado. Primore.) [49] A.edu/Icarus/.engin..) [56] D. (Quoted page 42.” http://www. Thirion.html. MN 55057.com.” Jun 2004.) [53] “SwissCube’s website. prepared for "SME" 2004. (Quoted page 10. com. “Design and implementation of on-board processor and software of student nanosatellite OUFTI-1 of University of Liège..Bibliography [43] P.) [55] P.ch/. revision J.” 2008.” http://swisscube. 3rd ed. sheldahl.) [46] F. 2007.” Master’s thesis. (Quoted pages 63 et 88. Indian Institute of Science. (Quoted page 47. VARTA PoLiFlex R Batteries Superior Polymer Technology. “Icarus Final Thermal Report. Bangalore.” http://www. Department of Instrumentation. Sheldahl. University of Liège. (Quoted page 42.” Laboratoire Leprince-Ringuet. com/online/cond_pcb. (Quoted page 42.frigprim.) [47] Pumpkin. Rochus. Rajendra. “Rosetta .

2003. 1999. T.) [60] P. 402. Rapport SF-258.sciencedirect. Microcosm Press and Kluwer Academic Publishers. Veughs and G.” Thermochimica Acta.” National Aeronautics And Space Administration. R. no. (Quoted page 21. Wiberg. patented Jul. and S. 1968. “ESARAD .” University of Liège. vol. Wertz and W.) [61] N. 1-2. (Quoted pages viii. Schmetz.” ABB Automation Technologies AB. “Specific heat capacity of lithium polymer battery components. Gramme Institute. Tech.) [62] J. pp. Introduction aux logiciels.ESATAN. Passerini. Space Mission Analysis and Desig. Rep. Carewska. Tech. 219 – 224. Belgium. [Online].Bibliography [59] P. Larson. Available: http://www. Villano.. “Conception et réalisation du système de déploiement des antennes du nanosatellite OUFTI-1. (Quoted page 47. Rep. Wakelyn.. 2009. “Thermal control coating. Rep. “Thermal study on hardware with comparisons between physical measurements and simulations.” Master’s thesis. J. Tech..) [64] A.) Lionel Jacques 130 2nd Master in Aerospace Engineering University of Liège Applied Sciences Faculty Academic Year 2008 2009 . 3rd ed. (Quoted pages ix et 60.com/science/article/ B6THV-47NF62S-4/2/4651f95f8ef619f68976a9d85dfe9b60 (Quoted page 59. may 2005. (Quoted page 42. May 2007. M. NASA. Belgium. Wertz. nov 1964.) [63] J. 9 et 70.