11.01.2015 Views

Master Thesis - OUFTI-1

Master Thesis - OUFTI-1

Master Thesis - OUFTI-1

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

University of Liège<br />

Faculty of Applied Sciences<br />

Aerospace and Mechanical Engineering Department<br />

<strong>Master</strong> <strong>Thesis</strong><br />

Dynamic Analysis and Launch Qualication of <strong>OUFTI</strong>-1<br />

Nanosatellite<br />

Nicolas FRANCOIS<br />

<strong>Thesis</strong> submitted in partial fulllment of the requirements for the degree of<br />

<strong>Master</strong> in Aerospace Engineering<br />

Advisor : Prof. Gaëtan KERSCHEN<br />

Academic Year 2009 - 2010


Acknowledgments<br />

I wish to acknowledge all the people who helped me, during the whole year, to carry<br />

out this master thesis.<br />

First of all, I would like to express my gratitude to my advisor Professor Gaëtan KER-<br />

SCHEN for his precious advice and support, and for giving me the opportunity to take<br />

one's rst steps in the fascinating world of space industries.<br />

I am deeply indebted to Professor Jean-Claude GOLINVAL who opened me the doors<br />

of his laboratory and gave me his condence to use his precious material.<br />

I would like to thank Professors Jean-Claude GOLINVAL and Pierre ROCHUS, but<br />

also Daniel SIMON and Amandine DENIS for accepting to serve on the examination commitee<br />

of this master thesis.<br />

I am pleased to acknowledge all the members of the <strong>OUFTI</strong> team, who created a pleasant<br />

working atmosphere. This year will remain a wonderful experience for me that I will<br />

never forget.<br />

Last but not least, I thank my family for encouraging and supporting me, not only<br />

during this academic year, but during my entire studies.<br />

1


Abstract<br />

<strong>OUFTI</strong>-1, standing for "Orbital Utility For Telecommunication Innovation ", is the rst<br />

nanosatellite developed by the University of Liège as well as the rst one which was ever<br />

developed in Belgium. This name, which is a typical expression from the city of Liège<br />

showing surprise or amazement, was chosen to point up the origin of the satellite. The<br />

project takes place within the framework of a long-term program called LEODIUM (which<br />

means Liège in Latin). The main goal of this program is to open a new window on scientic<br />

careers and space-related activities.<br />

<strong>OUFTI</strong>-1 will be equipped of three payloads:<br />

• A D-STAR repeater. <strong>OUFTI</strong>-1 will be the rst satellite to use this new digital<br />

amateur radio communication protocol in space.<br />

• A digitally-controlled electrical power system developed in collaboration with Thales<br />

Alenia Space ETCA.<br />

• High-eciency solar cells from Azur Space.<br />

<strong>OUFTI</strong>-1 will be launched on board the new european launcher: Vega.<br />

This present thesis focuses on the structural design and dynamic analysis of <strong>OUFTI</strong>-<br />

1. The design part covers essentially the creation of a reliable support for the batteries<br />

in accordance with several requirements. It includes also some changes in the general<br />

conguration of the satellite. The analysis part is based on the creation of an accurate<br />

nite element modeling procedure for electronic cards. Then, this procedure is applied<br />

to each electronic card and several FE analysis are performed to ensure the structural<br />

integrity of the satellite. Finally, a random vibration analysis is presented. The goal of<br />

this analysis is to demonstrate the ability of <strong>OUFTI</strong>-1 to survive at the launch phase.<br />

2


Contents<br />

1 Introduction 13<br />

1.1 The CubeSat program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13<br />

1.1.1 The concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13<br />

1.1.2 The Pumpkin's CubeSat kit . . . . . . . . . . . . . . . . . . . . . . 14<br />

1.1.3 The P-POD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15<br />

1.2 <strong>OUFTI</strong>-1 project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16<br />

1.2.1 Genesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16<br />

1.2.2 The launch opportunity . . . . . . . . . . . . . . . . . . . . . . . . 16<br />

1.2.3 Mission objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18<br />

1.2.4 <strong>OUFTI</strong> team . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19<br />

2 <strong>OUFTI</strong>-1: Flight system conguration and general properties 20<br />

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20<br />

2.2 Reference frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20<br />

2.3 Flight system conguration . . . . . . . . . . . . . . . . . . . . . . . . . . 21<br />

2.3.1 Skeleton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23<br />

2.3.2 Solar panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24<br />

2.3.3 Antenna support . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26<br />

2.3.4 Permanent magnet and hysteretic bars . . . . . . . . . . . . . . . . 27<br />

2.3.5 Electronic cards and battery support . . . . . . . . . . . . . . . . . 32<br />

2.4 Physical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35<br />

2.4.1 Catia modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35<br />

2.4.2 Center of gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35<br />

2.4.3 Inertia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37<br />

2.5 Mass budget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38<br />

2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39<br />

3 Design of a new support for the batteries 44<br />

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44<br />

3.2 Problems encountered with the last year's design . . . . . . . . . . . . . . 44<br />

3.3 Available mass budget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46<br />

3


3.4 Initial idea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47<br />

3.5 Battery selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49<br />

3.6 Material selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51<br />

3.6.1 Denition of the requirements . . . . . . . . . . . . . . . . . . . . . 51<br />

3.6.2 Objective function and CES software . . . . . . . . . . . . . . . . . 53<br />

3.7 Thermal control's devices selection . . . . . . . . . . . . . . . . . . . . . . 55<br />

3.7.1 Heaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55<br />

3.7.2 Heaters' control system . . . . . . . . . . . . . . . . . . . . . . . . . 56<br />

3.7.3 Thermal insulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58<br />

3.8 Final design of the support . . . . . . . . . . . . . . . . . . . . . . . . . . . 59<br />

3.9 Validation of the design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63<br />

3.9.1 Mass budget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63<br />

3.9.2 Static loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65<br />

3.9.3 Dynamic behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . 68<br />

3.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70<br />

4 Electronic cards dynamic modeling 72<br />

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72<br />

4.2 State-of-the-art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72<br />

4.2.1 Accuracy of the model . . . . . . . . . . . . . . . . . . . . . . . . . 73<br />

4.2.2 Creating FE models of PCBs . . . . . . . . . . . . . . . . . . . . . 74<br />

4.2.3 Determination of the PCB properties . . . . . . . . . . . . . . . . . 74<br />

4.2.4 Choice of mesh element . . . . . . . . . . . . . . . . . . . . . . . . . 76<br />

4.2.5 Recognition of components eects . . . . . . . . . . . . . . . . . . . 79<br />

4.2.6 Modeling of the chassis . . . . . . . . . . . . . . . . . . . . . . . . . 82<br />

4.2.7 Denition of the boundary conditions . . . . . . . . . . . . . . . . . 82<br />

4.2.8 Introduction of damping . . . . . . . . . . . . . . . . . . . . . . . . 82<br />

4.2.9 Dynamic response computation . . . . . . . . . . . . . . . . . . . . 84<br />

4.3 Application to the homemade on-board computer of <strong>OUFTI</strong>-1 . . . . . . . 85<br />

4.3.1 Denition of the PCB properties . . . . . . . . . . . . . . . . . . . 85<br />

4.3.2 Recognition of the components eects . . . . . . . . . . . . . . . . . 86<br />

4.3.3 Experimental test of the OBC 2 card . . . . . . . . . . . . . . . . . 88<br />

4.3.4 Application of the simple method . . . . . . . . . . . . . . . . . . . 91<br />

4.3.5 Application of the global mass smearing method . . . . . . . . . . . 93<br />

4.3.6 Application of the local mass smearing method . . . . . . . . . . . 94<br />

4.3.7 Application of a homemade method . . . . . . . . . . . . . . . . . . 95<br />

4.3.8 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 97<br />

4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100<br />

4


5 Complete FE analysis of <strong>OUFTI</strong>-1: Static, modal and random vibration101<br />

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101<br />

5.2 Modeling strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101<br />

5.2.1 Materials properties . . . . . . . . . . . . . . . . . . . . . . . . . . 102<br />

5.2.2 Main structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102<br />

5.2.3 Solar panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103<br />

5.2.4 Antenna support . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104<br />

5.2.5 Electronic cards and battery support . . . . . . . . . . . . . . . . . 104<br />

5.2.6 The endless screws . . . . . . . . . . . . . . . . . . . . . . . . . . . 108<br />

5.2.7 ADCS and THER components . . . . . . . . . . . . . . . . . . . . . 109<br />

5.3 Static analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110<br />

5.3.1 Loads denition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110<br />

5.3.2 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . 112<br />

5.3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112<br />

5.4 Modal analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115<br />

5.4.1 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115<br />

5.5 Random vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117<br />

5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119<br />

6 Conclusions 120<br />

6.1 Summary of the accomplished work . . . . . . . . . . . . . . . . . . . . . . 120<br />

6.2 Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122<br />

6.3 Parallel activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123<br />

5


List of Figures<br />

1.1 The Pumpkin's CubeSat family . . . . . . . . . . . . . . . . . . . . . . . . 15<br />

1.2 Poly-Picosatellite Orbital Deployer (P-POD) . . . . . . . . . . . . . . . . . 15<br />

1.3 The new European launcher: Vega . . . . . . . . . . . . . . . . . . . . . . 16<br />

1.4 Main stage of Vega [20] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17<br />

1.5 The three payloads of <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . . . . . . . . 18<br />

2.1 Reference frame of <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . . . . . . . . . . 21<br />

2.2 Product tree of <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22<br />

2.3 CSK structure with the FM 430 . . . . . . . . . . . . . . . . . . . . . . . . 23<br />

2.4 Dose depth curve for <strong>OUFTI</strong>-1 orbit [6] . . . . . . . . . . . . . . . . . . . . 24<br />

2.5 One fully-integrated solar panel of <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . 25<br />

2.6 Available holes on the CSK structure for cabling path [15] . . . . . . . . . 25<br />

2.7 Interference between solar panel clips and <strong>OUFTI</strong>-1 solar cells [15] . . . . . 26<br />

2.8 Conguration and dimensions of the antenna support . . . . . . . . . . . . 27<br />

2.9 Description of the port obstructed by the antenna support . . . . . . . . . 27<br />

2.10 Principle of the <strong>OUFTI</strong>-1 ADCS subsystem [25] . . . . . . . . . . . . . . . 28<br />

2.11 Last year's conguration of the ADCS . . . . . . . . . . . . . . . . . . . . 28<br />

2.12 Ideal positioning of the hysteretic bars . . . . . . . . . . . . . . . . . . . . 29<br />

2.13 Magnetic eld obtained in the hysteretic bars for the rst conguration [25] 30<br />

2.14 Magnetic eld obtained in the hysteretic bars for the second conguration<br />

[25] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30<br />

2.15 Final conguration of the ADCS . . . . . . . . . . . . . . . . . . . . . . . . 31<br />

2.16 Magnetic eld obtained in the hysteretic bars for the nal conguration [25] 31<br />

2.17 Pictures of the ADCS components [25] . . . . . . . . . . . . . . . . . . . . 32<br />

2.18 Conguration of <strong>OUFTI</strong>-1 electronic cards . . . . . . . . . . . . . . . . . . 33<br />

2.19 Pictures of the engineering models of <strong>OUFTI</strong>-1 electronic cards . . . . . . 34<br />

2.20 Exploded view of <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . . . . . . . . . . 36<br />

3.1 Last year's design of the battery assembly . . . . . . . . . . . . . . . . . . 44<br />

3.2 Available space under and above the batteries . . . . . . . . . . . . . . . . 45<br />

3.3 Batteries during and after the vacuum test [15] . . . . . . . . . . . . . . . 46<br />

3.4 Initial idea for the new design . . . . . . . . . . . . . . . . . . . . . . . . . 48<br />

6


3.5 Adaptation of the base idea to the <strong>OUFTI</strong>-1 dimensions . . . . . . . . . . 49<br />

3.6 Available space for including the batteries . . . . . . . . . . . . . . . . . . 49<br />

3.7 Battery Kokam SLB 603870H . . . . . . . . . . . . . . . . . . . . . . . . . 50<br />

3.8 Classication by density and Young's modulus . . . . . . . . . . . . . . . . 53<br />

3.9 Classication by density and yield stress . . . . . . . . . . . . . . . . . . . 54<br />

3.10 Polyimide thermofoil heater . . . . . . . . . . . . . . . . . . . . . . . . . . 55<br />

3.11 Thermostat Klixon 4BT − 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 56<br />

3.12 Connection with four thermostats . . . . . . . . . . . . . . . . . . . . . . . 57<br />

3.13 Thermal insulator: Polyester netting . . . . . . . . . . . . . . . . . . . . . 58<br />

3.14 Thickness of the box's "base plate" . . . . . . . . . . . . . . . . . . . . . . 59<br />

3.15 P C 104 connector's hole . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59<br />

3.16 Reinforcers and connection holes . . . . . . . . . . . . . . . . . . . . . . . . 60<br />

3.17 Final design of the box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60<br />

3.18 Schematics of the box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61<br />

3.19 Reinforcements around the connection holes . . . . . . . . . . . . . . . . . 61<br />

3.20 Notches for the thermostats . . . . . . . . . . . . . . . . . . . . . . . . . . 62<br />

3.21 End stops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62<br />

3.22 Final design of the cover . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63<br />

3.23 Schematics of the cover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64<br />

3.24 Maximal Von Mises stress inside the support . . . . . . . . . . . . . . . . . 67<br />

3.25 The two most signicant modes of the battery support . . . . . . . . . . . 69<br />

3.26 A rst rough shape of the battery support . . . . . . . . . . . . . . . . . . 70<br />

4.1 Example of manufacturing variation [44] . . . . . . . . . . . . . . . . . . . 74<br />

4.2 Example of dierences between the values provided by manufacturers and<br />

the values determined by experimental tests [44] . . . . . . . . . . . . . . . 75<br />

4.3 Illustration of the experimental set-up for a static bend test . . . . . . . . 75<br />

4.4 Illustration of the experimental set-up for a torsion test [44] . . . . . . . . 76<br />

4.5 MAC between the two models . . . . . . . . . . . . . . . . . . . . . . . . . 78<br />

4.6 Symmetrical modes of the steel plate . . . . . . . . . . . . . . . . . . . . . 78<br />

4.7 Detailed model of a component [45] . . . . . . . . . . . . . . . . . . . . . . 79<br />

4.8 Illustration of the dierent simplication methods [45] . . . . . . . . . . . . 80<br />

4.9 Illustration of the dierent categories of components [45] . . . . . . . . . . 81<br />

4.10 Illustration of the peak-amplitude method [53] . . . . . . . . . . . . . . . . 83<br />

4.11 Location of the OBC 2 card in <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . . . 85<br />

4.12 Components classication of the OBC 2 card . . . . . . . . . . . . . . . . . 86<br />

4.13 Catia modeling of the OBC 2 card . . . . . . . . . . . . . . . . . . . . . . 87<br />

4.14 Example of spacecraft stiness requirements for dierent launchers . . . . . 88<br />

4.15 Location of measurement and excitation points . . . . . . . . . . . . . . . . 90<br />

4.16 Experimental set-up for the test of the OBC 2 card . . . . . . . . . . . . . 90<br />

4.17 Global FRF of the OBC 2 card . . . . . . . . . . . . . . . . . . . . . . . . 91<br />

7


4.18 MAC matrix using the simple method . . . . . . . . . . . . . . . . . . . . 92<br />

4.19 MAC matrix using the global mass smearing method . . . . . . . . . . . . 93<br />

4.20 MAC matrix using the local mass smearing method . . . . . . . . . . . . . 95<br />

4.21 MAC matrix using the homemade method . . . . . . . . . . . . . . . . . . 96<br />

4.22 Initial shape of the OBC 2 card's fourth mode . . . . . . . . . . . . . . . . 97<br />

4.23 Mode shapes obtained by adding sensors of several mass to the FE model . 98<br />

4.24 Inuence of a change of modeling strategy for the P C 104 connector on the<br />

OBC 2 card's fourth mode . . . . . . . . . . . . . . . . . . . . . . . . . . . 99<br />

5.1 Illustration of a "glue" assembly [56] . . . . . . . . . . . . . . . . . . . . . 103<br />

5.2 The FM 430 card from Pumpkin . . . . . . . . . . . . . . . . . . . . . . . 105<br />

5.3 The OBC 2 card . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105<br />

5.4 The EPS card . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106<br />

5.5 The xEPS card developed in collaboration with Thales Alenia Space ETCA 107<br />

5.6 Modeling of the midplane standos . . . . . . . . . . . . . . . . . . . . . . 109<br />

5.7 Illustration of the worst case conguration . . . . . . . . . . . . . . . . . . 110<br />

5.8 Frame rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111<br />

5.9 Illustration of the complete quasi-static loading state of <strong>OUFTI</strong>-1 . . . . . 112<br />

5.10 Displacements of <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 113<br />

5.11 Maximal Von Mises stress inside <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . . 114<br />

5.12 First mode of <strong>OUFTI</strong>-1 (325.51 Hz) . . . . . . . . . . . . . . . . . . . . . 116<br />

5.13 Second mode of <strong>OUFTI</strong>-1 (333.92 Hz) . . . . . . . . . . . . . . . . . . . . 116<br />

5.14 PSD of several LVs at qualication level [60] . . . . . . . . . . . . . . . . . 117<br />

5.15 Precise PSD of Vega [57] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118<br />

6.1 Schemas of solar panels +X and +/ − Y [15] . . . . . . . . . . . . . . . . . 130<br />

6.2 Schemas of the solar panel +Z [15] . . . . . . . . . . . . . . . . . . . . . . 131<br />

6.3 Schemas of the solar panel −Z [15] . . . . . . . . . . . . . . . . . . . . . . 131<br />

6.4 Schemas of electronic cards [15] . . . . . . . . . . . . . . . . . . . . . . . . 132<br />

6.5 Available volume for each electronic card [62] . . . . . . . . . . . . . . . . . 133<br />

6.6 Fixations of endless screws on the base plate [62] . . . . . . . . . . . . . . 134<br />

6.7 Picture of the xation between one midplane stando and the chassis [33] . 135<br />

6.8 Disposition of midplane standos inside the CubeSat [30] . . . . . . . . . . 135<br />

6.9 Datasheet of the battery Kokam SLB 603870H . . . . . . . . . . . . . . . 137<br />

8


List of Tables<br />

2.1 Coordinates of the <strong>OUFTI</strong>-1 CoG . . . . . . . . . . . . . . . . . . . . . . . 37<br />

2.2 Mass budget of <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43<br />

3.1 Available mass budget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47<br />

3.2 Important corrections in the mass budget . . . . . . . . . . . . . . . . . . . 47<br />

3.3 Properties of several batteries from Kokam [36] . . . . . . . . . . . . . . . 50<br />

3.4 Materials properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52<br />

3.5 Dierent available congurations and problems for thermostats . . . . . . . 58<br />

3.6 Necessary mass budget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65<br />

3.7 First natural frequencies of the battery support . . . . . . . . . . . . . . . 68<br />

4.1 Frequency deviations between the two models . . . . . . . . . . . . . . . . 77<br />

4.2 Properties of the FR 4 material . . . . . . . . . . . . . . . . . . . . . . . . 85<br />

4.3 Frequency deviations between corresponding modes using the simple method 91<br />

4.4 Frequency deviations between corresponding modes using the global mass<br />

smearing method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93<br />

4.5 Increases of density due to each component of the OBC 2 card . . . . . . . 94<br />

4.6 Frequency deviations between corresponding modes using the local mass<br />

smearing method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94<br />

4.7 Frequency deviations between corresponding modes using the homemade<br />

method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96<br />

4.8 Natural frequencies obtained by adding sensors of several mass to the FE<br />

model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98<br />

5.1 Properties of the materials used in <strong>OUFTI</strong>-1 . . . . . . . . . . . . . . . . . 102<br />

5.2 Properties of the FM 430 card . . . . . . . . . . . . . . . . . . . . . . . . . 105<br />

5.3 Properties of the OBC 2 card . . . . . . . . . . . . . . . . . . . . . . . . . 106<br />

5.4 Properties of the EPS card . . . . . . . . . . . . . . . . . . . . . . . . . . . 106<br />

5.5 Properties of the xEPS card . . . . . . . . . . . . . . . . . . . . . . . . . . 107<br />

5.6 Properties of the COM card . . . . . . . . . . . . . . . . . . . . . . . . . . 108<br />

5.7 Results of the static analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 112<br />

5.8 Five rst natural frequencies of <strong>OUFTI</strong>-1 in hard-mounted conguration . 115<br />

9


List of Acronyms<br />

ADCS<br />

Al<br />

AlNiCo<br />

ASI<br />

BGA<br />

CAD<br />

CalPoly<br />

CDS<br />

CFRP<br />

CoG<br />

COM<br />

COTS<br />

CSL<br />

CSK<br />

CTE<br />

CVCM<br />

D-STAR<br />

DAR<br />

DoF<br />

EPS<br />

ESA<br />

ESTEC<br />

FE<br />

FM<br />

FR<br />

FRF<br />

GaAs<br />

GFRP<br />

GPS<br />

ICD<br />

IRM<br />

Attitude Determination and Control System<br />

Aluminum<br />

Aluminum-Nickel-Cobalt<br />

Italian Space Agency<br />

Ball Grid Array<br />

Computer-Aided Design<br />

California Polytechnic State University<br />

CubeSat Design Specication<br />

Carbon Fiber Reinforced Plastic<br />

Center of Gravity<br />

Telecommunication<br />

Commercial-O-The-Shelf<br />

Liège Space Center<br />

CubeSat Kit<br />

Coecient of Thermal Expansion<br />

Collected Volatile Condensable Material<br />

Digital Smart Technology for Amateur Radio<br />

Deviation Waiver Approval Request<br />

Degree of Freedom<br />

Electrical Power Supply<br />

European Space Agency<br />

European Space Research and Technology Center<br />

Finite Element<br />

Flight Module<br />

Flame Resistant<br />

Frequency Response Function<br />

Gallium-Arsenide<br />

Glass Fiber Reinforced Plastic<br />

Global Positioning System<br />

Interface Control Document<br />

Royal Meteorological Institute<br />

10


ITD Ibrahim Time Domain<br />

ITR Interface Technical Review<br />

JARL Japan Amateur Radio League<br />

KISS Keep It Simple, Stupid (and Short)<br />

LEODIUM Low Earth Orbit Demonstration In University Mode<br />

Li Lithium<br />

LV Launch Vehicle<br />

MAC Modal Assurance Criterion<br />

MECH Mechanism<br />

MLI Multilayer Insulation<br />

MoS Margin of Safety<br />

OBC On-Board Computer<br />

OBC 1 Pumpkin's On-Board computer (FM 430)<br />

OBC 2 Homemade On-Board Computer<br />

<strong>OUFTI</strong> Orbital Utility For Telecommunication Innovation<br />

P-POD Poly-Picosatellite Orbital Deployer<br />

PCB Printed Circuit Board<br />

PGA Pin Grid Array<br />

PSD Power Spectral Density<br />

QFP Quad Flat Pack<br />

RBF Remove Before Flight<br />

RF Radio Frequency<br />

SDR Software-Dened Radio<br />

SF Safety Factor<br />

SMT Surface Mount Technology<br />

SSI Stochastic Subspace Identication<br />

STRU Structure & Conguration<br />

THER Thermal Control<br />

T i Titanium<br />

TID Total Irradiation Dose<br />

TML Total Mass Loss<br />

UCL Catholic University of Louvain-la-Neuve<br />

UHF Ultra High Frequency<br />

UHMWPE Ultra High Molecular Weight Polyethylene<br />

ULg University of Liège<br />

VCD Verication Control Document<br />

VHF Very High Frequency<br />

xEPS Experimental Electrical Power Supply<br />

11


<strong>Thesis</strong> outline<br />

This thesis focuses on the structural design of a new support for the batteries and dynamic<br />

analysis of the nanosatellite <strong>OUFTI</strong>-1, developed at the University of Liège (ULg).<br />

This work will be divided into 6 parts:<br />

• Firstly, a brief introduction about the project and its history will be presented.<br />

• Then, the general conguration of the satellite will be discussed. The requirements<br />

and constraints which led to the nal conguration, will be introduced subsystem by<br />

subsystem to bring an overall view of the project.<br />

• In the third chapter, the design of a new support for the batteries will be studied.<br />

The reasons that led to change the support created last year, will be exposed and<br />

the development phase will be discussed step by step.<br />

• After this structural design part, we will focus on dynamic analysis. The rst step<br />

will be to create a dynamic modeling procedure for the electronic cards. The dierent<br />

methods that already exist to realize an accurate Finite Element (FE) model will be<br />

discussed. Then, the best one will be applied to the Homemade On-Board Computer<br />

(OBC 2) of <strong>OUFTI</strong>-1.<br />

• Applying this procedure to each electronic card of <strong>OUFTI</strong>-1, complete static and<br />

dynamic analysis will be performed. Random vibration analysis will also be presented<br />

to demonstrate the structural integrity of the satellite during the launch phase.<br />

• Finally, the conclusions of this thesis will be drawn and perspectives for future works<br />

will be highlighted.<br />

12


Chapter 1<br />

Introduction<br />

1.1 The CubeSat program<br />

1.1.1 The concept<br />

The <strong>OUFTI</strong>-1 project takes place in the CubeSat program initiated approximately 10<br />

years ago jointly by California Polytechnic State University (CalPoly) and Stanford University.<br />

This program aims to provide an easier access to space by creating a standard based<br />

on the following features:<br />

• The mass of the satellite must be less than 1 kg (1.33 kg in the last version of the<br />

CubeSat Design Specication (CDS) [1], but this version stands not for the Vega<br />

Maiden Flight on board which <strong>OUFTI</strong>-1 will be launched).<br />

• The shape of the satellite must be a cube of 10 × 10 × 10 cm.<br />

This standard allows to create satellites in a short development time. In addition, owing<br />

to their small size, CubeSats can take place in a space launcher as secondary payloads,<br />

which is less expensive. For all of this reasons, the CubeSat program meets a great success<br />

with universities and industries.<br />

At the moment, the project involves more than 100 universities, high schools or private<br />

rms all over the world developing CubeSats containing scientic, private and governments<br />

payloads. All these developers are linked together by a great community spirit, and each<br />

one benets from the sharing of information between the several teams [1].<br />

However, it is important to note that, owing to constraints listed above, the development<br />

of a CubeSat is a technical challenge, even if it is faster and cheaper than the development<br />

13


of larger satellites. So, the CubeSat philosophy is quite dierent. Indeed, the leitmotiv of<br />

several CubeSat teams is KISS: "Keep It Simple, Stupid (and Short)".<br />

This philosophy is based on the fact that, if two dierent solutions are available to<br />

realize the same task with the same eciency, the best one is the simplest one.<br />

This explains that CubeSats projects use Commercial-O-The-Shelf (COTS) components<br />

instead of space-qualied ones. It is a particularity of these projects and, for this<br />

reason, the ocial requirements must be lightened for this type of mission.<br />

The denition provided here is a general one. All the technical requirements that a<br />

nanosatellite has to respect to be validated as a CubeSat are available in reference [1].<br />

1.1.2 The Pumpkin's CubeSat kit<br />

In order to answer to the request which grows incessantly, some industries have decided<br />

to develop structures that respect these requirements. Among all these commercial structures,<br />

the most popular remains the CubeSat Kit (CSK) from Pumpkin [2]. This structure<br />

is available in three dierent formats (as shown in Figure 1.1).<br />

For <strong>OUFTI</strong>-1, the goal was to create a 1 unit CubeSat. So, after a long hesitation<br />

between the Pumpkin's structure and an other one provided by ISIS, it was decided to buy<br />

the 1 unit skeleton from Pumpkin.<br />

The decision of buying a commercial structure was chosen because the development<br />

of a space structure is a complicated problem. Indeed, requirements imposed to spacequalied<br />

devices are so restrictive that the development of a space-qualied structure is<br />

too ambitious for ULg, which had absolutely no experience in satellite development before<br />

this project.<br />

14


1.1.3 The P-POD<br />

Figure 1.1: The Pumpkin's CubeSat family<br />

The P-POD, which stands for "Poly-Picosatellite Orbital Deployer ", is the standardized<br />

CubeSat deployment system developed by CalPoly (shown in Figure 1.2). It is a rectangular<br />

box with a door (situated on the right side in Figure 1.2). It serves as the interface between<br />

CubeSats and the Launch Vehicle (LV). Three CubeSats can be placed within one P-POD.<br />

When the launcher is in orbit, the P-PODs deploy the CubeSats using a spring mechanism<br />

to push CubeSats along a series of rails and eject them into orbit. Further detail can be<br />

found in reference [3].<br />

Figure 1.2: Poly-Picosatellite Orbital Deployer (P-POD)<br />

15


1.2 <strong>OUFTI</strong>-1 project<br />

1.2.1 Genesis<br />

<strong>OUFTI</strong>-1, standing for "Orbital Utility For Telecommunication Innovation ", is the rst<br />

nanosatellite developed by the University of Liège (ULg) as well as the rst one which was<br />

ever developed in Belgium. This name, which is a typical expression from the city of Liège<br />

showing surprise or amazement, was chosen to point up the origin of the satellite.<br />

The project started in 2007 when Mr. Luc HALBACH, an engineer of Spacebel, suggests<br />

the idea of testing a new amateur radio digital technology in space on board a CubeSat: the<br />

D-STAR protocol. This idea has catched the attention of several academic members from<br />

ULg and, within a few weeks, a team was created to work on the project: <strong>OUFTI</strong>-1 was<br />

born. This team includes two students from ULg, Stefania GALLI and Jonathan PISANE,<br />

who realized the mission design of <strong>OUFTI</strong>-1 [4] and the design and implementation of the<br />

rst telecommunication elements of <strong>OUFTI</strong>-1 [5] respectively. Last year, it was a great<br />

team composed of 13 students coming from three dierent schools (University of Liège,<br />

Gramme Institute and ISIL Institute), who developed the project to the state at which<br />

this thesis starts (references [6] to [18]).<br />

1.2.2 The launch opportunity<br />

<strong>OUFTI</strong>-1 will be launched on board the new european launcher: Vega (shown in Figure<br />

1.3) for free, which is an incredible opportunity. This launcher was developed by Arianespace<br />

jointly to the Italian Space Agency (ASI) and European Space Agency (ESA). Further<br />

detail can be found in reference [19]. The main payload of the Vega Maiden Flight is the<br />

LARES system (an articial satellite provided by ASI with a spherical body and covered<br />

by corner cube reectors). The secondary payloads includes 9 CubeSats, plus the satellite<br />

ALMASat-1 from the University of Bologna.<br />

Figure 1.3: The new European launcher: Vega<br />

16


Figure 1.4 shows the main stage of Vega during a vibration test. The LARES system<br />

is situated on the top of the stage and, on the left of this picture, one P-POD can be seen.<br />

Last year, the P-PODs were theoretically placed horizontally inside the launcher. However,<br />

in order to facilitate the access to them (e.g., for recharging the batteries), it was decided<br />

to right them at an angle of 10 ◦ from the vertical. This fact is really important to note<br />

because this modication will change signicantly the loads apply on the CubeSat.<br />

Figure 1.4: Main stage of Vega [20]<br />

This opportunity results from an initiative of ESA that, in January 2008, published a<br />

call to oer a free launch for nine CubeSats on the Vega Maiden Flight after presenting the<br />

project at the Vega Maiden Flight CubeSat Workshop at the European Space Research<br />

and Technology Center (ESTEC). The <strong>OUFTI</strong>-1 team has submitted its own proposal in<br />

March 2008 and, in June 2008, received the conrmation that <strong>OUFTI</strong>-1 will take part in<br />

the Vega Maiden Flight. The eight other selected CubeSats and their specic missions,<br />

are:<br />

• Robusta (Montpellier - France): Study of radiation eects on bipolar transistors<br />

• UWE-3 (Würzburg - Germany): Development of an active attitude control system<br />

• AtmoCube (Trieste - Italy): Space weather measurements<br />

• E-St@r (Torino - Italy): Test of an active 3-axis attitude control system<br />

• UNICubeSat (Roma - Italy): Atmospheric neutral density measurements<br />

• PW-Sat (Warsaw - Poland): Test of a deployable drag augmentation device<br />

• Goliat (Bucharest - Romania): Earth imaging and space environment measurements<br />

17


• XatCobeo (Vigo - Spain): Development of a Software-Dened Radio (SDR) and solar<br />

panel deployment system<br />

The Vega Maiden Flight will deploy these CubeSats into an orbit of 354 × 1447 km<br />

altitude with 71 ◦ inclination.<br />

1.2.3 Mission objectives<br />

The primary goal of the <strong>OUFTI</strong>-1 project is to provide hands-on satellite experience to<br />

students. For the <strong>OUFTI</strong>-1 project, there are three main payloads (illustrated in Figure<br />

1.5).<br />

Figure 1.5: The three payloads of <strong>OUFTI</strong>-1<br />

The rst one is the setting up of a functional D-STAR repeater in space. D-STAR,<br />

which stands for "Digital Smart Technology for Amateur Radio ", is a ham radio protocol<br />

recently developed by the Japan Amateur Radio League (JARL). The overall system<br />

provides a lot of new built-in features including digital communication (i.e., the quality<br />

of the data received is better than an analog signal at the same strength), simultaneous<br />

voice and data transmission (e.g., Global Positioning System (GPS) data and computer<br />

les), complete routing over the Internet and callsign-based roaming on a worldwide basis<br />

[21]. Therefore, the D-STAR system provides a new capability and functionality to the<br />

ham radio world and increases the eciency of emergency communications (e.g., during<br />

the hurricane Katrina, the whole telecommunications were out of order and it is amateur<br />

radio that allowed people to communicate with the rescue teams). This payload allows to<br />

test the performances of this protocol in space environment.<br />

18


The second one is an Experimental Electrical Power Supply (xEPS) developed in collaboration<br />

with Thales Alenia Space ETCA. xEPS is based on a digitally-controlled yback<br />

converter [17]. This type of electrical power system was never tested in space environment<br />

so, to avoid any problem, this system is redundant with an analogical one (which is commonly<br />

used in space applications).<br />

The third payload is high-eciency solar cells provided by Azur Space. This new generation<br />

of solar cells is triple-junction Gallium-Arsenide (GaAs) cells with an eciency of<br />

30%.<br />

These two last payloads result from the fact that nanosatellites are ideal low-cost solutions<br />

for testing new technologies in space environment. So, industries present a real<br />

interest for these missions and could provide some funds or equipments just to prove that<br />

their technology is ecient.<br />

1.2.4 <strong>OUFTI</strong> team<br />

This year, the team is composed of two project managers, a system engineering team<br />

composed of three graduate students, most of them being involved in the project last<br />

year, several professors and 14 second master students from dierent institutions (ULg,<br />

Gramme Institute, ISIL Institute and Catholic University of Louvain-la-Neuve (UCL)).<br />

Most of these students are directly involved in the <strong>OUFTI</strong>-1 project but some of them<br />

alredy work on the future projects.<br />

19


Chapter 2<br />

<strong>OUFTI</strong>-1: Flight system conguration<br />

and general properties<br />

2.1 Introduction<br />

In consequence of the restricted dimensions of a CubeSat, the management of the available<br />

space to integrate all the components of each subsystem is a real challenge.<br />

Through this chapter, we will focus on the organization of subsystems inside the satellite.<br />

For each of them, the requirements that led to this particular choice, and the progression<br />

from the starting point [15] to the nal conguration, will be presented.<br />

An other objective is to bring an overall view of the satellite, which will allow to have<br />

a better understanding of the following chapters.<br />

2.2 Reference frame<br />

First of all, a reference frame must be dened. Its origin is located at the CubeSat<br />

geometric center. The frame is oriented using the right-hand rule (as shown in Figure 2.1).<br />

This particular frame presents 3 major advantages:<br />

• It is the same frame as the one given in the ocial documents [1].<br />

• The distance, that separates the Center of Gravity (CoG) from the geometric center,<br />

can be directly evaluated.<br />

• It allows to locate each component of the satellite in an easy way. For example,<br />

the faces are identied using the name of the axis which is perpendicular to them<br />

20


Figure 2.1: Reference frame of <strong>OUFTI</strong>-1<br />

preceded by the direction of this axis (e.g., the face where the access port area is<br />

located, is referred as face −X).<br />

2.3 Flight system conguration<br />

Through this section, the general conguration of the satellite will be discussed. Each<br />

component will be reviewed in a global understanding approach. Further detail on each of<br />

them can be found in reference [15].<br />

First of all, the product tree of the entire satellite is presented in Figure 2.2. This<br />

one allows to have an overall view of how the several components can be sorted. Indeed,<br />

<strong>OUFTI</strong>-1 can be divided into three main categories:<br />

• The main structure, including all CSK structural parts, and components which are<br />

connected to external faces of the <strong>OUFTI</strong>-1 skeleton.<br />

• The internal equipments, including all electronic cards and structural devices which<br />

are located inside the satellite.<br />

• The xations, including all screws and other xing means which allow to connect all<br />

<strong>OUFTI</strong>-1 components together.<br />

21


Figure 2.2: Product tree of <strong>OUFTI</strong>-1<br />

22


2.3.1 Skeleton<br />

The skeleton of <strong>OUFTI</strong>-1 corresponds to the 1U structure of Pumpkin's CSK. This<br />

structure comprises three main parts:<br />

• A chassis, with a thickness of 1.27 mm.<br />

• A base plate, with a thickness of 1.52 mm, that is screwed on the chassis with 6<br />

M3 × 4 mm stainless steel screws. It supports one normal foot, two spring plunger<br />

feet and one deployment switch foot. These special feet are used to deploy the<br />

CubeSats from the P-PODs when they are placed in orbit.<br />

• An end plate, with a thickness of 1.52 mm, that is screwed on the chassis with 4<br />

M3 × 4 mm stainless steel screws. It supports four normal feet.<br />

The skeleton is made of Aluminum (Al) 5052 H32 and the feet are made of Al−6061 T 6.<br />

A surface treatment is also performed on the structure. The rails (situated on the corners<br />

of the chassis) and the feet are hard-anodized to prevent galling when they are in contact<br />

with the P-POD. The rest of the structure is alodined, which is an electrically-conductive<br />

surface nish [22]. This last fact is very important because the structure will be used as<br />

the reference electrical mass for the CubeSat.<br />

The kit also includes a Flight Module (FM) 430 (used as main On-Board Computer<br />

(OBC) in <strong>OUFTI</strong>-1: OBC 1). It is important to note that all the functions available on<br />

this card are not used for <strong>OUFTI</strong>-1 (e.g., the MHX transceiver). This fact will turn out to<br />

be a crucial information for the nal conguration (see section 2.3.3).<br />

Figure 2.3 shows the devices of the CubeSat kit bought by ULg.<br />

Figure 2.3: CSK structure with the FM 430<br />

23


2.3.2 Solar panels<br />

The armor panels of <strong>OUFTI</strong>-1 are made of Al − 7075 T 73. Their role is to protect<br />

all the equipments of the satellite against the harsh space environment: cosmic radiations,<br />

debris, thermal variations, ...<br />

Last year, a simulation of the <strong>OUFTI</strong>-1 orbit was performed to obtain the Total Irradiation<br />

Dose (TID) at which the satellite is submitted during one year [6]. The TID is<br />

the addition of the eects of the trapped electrons and protons within the radiation belts,<br />

the solar protons and the Bremsstrahlung eect. The results are presented in Figure 2.4,<br />

where the TID is considered as a function of equivalent aluminum shielding thickness.<br />

Figure 2.4: Dose depth curve for <strong>OUFTI</strong>-1 orbit [6]<br />

In the case of <strong>OUFTI</strong>-1, the thickness of these panels was chosen to 1.5 mm, which corresponds<br />

to an equivalent aluminum shielding thickness of 1.95 mm for the entire satellite.<br />

The complete study that led to this particular choice is available in reference [15]. There<br />

are ve panels mounted on the <strong>OUFTI</strong>-1 skeleton:<br />

• Two panels, one on the base plate and the other on the end plate, with external<br />

dimensions of 90 × 96 mm 2 .<br />

• Three panels placed on the faces +X and +/ − Y of the chassis, with external<br />

dimensions of 82 × 96 mm 2 24


There is no solar panel on the face −X because the access port area can not be obstructed.<br />

The solar cells were mounted on these panels at the rate of two per panel. Their integration<br />

was performed by EADS Astrium. A picture of a fully-integrated panel is available<br />

in Figure 2.5. Schematics and exact dimensions of each of these panels are available in<br />

appendix A.<br />

Figure 2.5: One fully-integrated solar panel of <strong>OUFTI</strong>-1<br />

One of the problem encountered is that these panels cover integrally the holes of the<br />

faces on which they are placed. So, to provide a way for the cabling path, some of them<br />

are drilled of several holes with a diameter of 3 mm. The available holes for the cabling<br />

path are presented in Figure 2.6.<br />

Figure 2.6: Available holes on the CSK structure for cabling path [15]<br />

Finally, it should be noted that these panels will be glued on the <strong>OUFTI</strong>-1 skeleton<br />

using a specic glue: epoxy Stycast 2850 (with catalyst 24 LV), chosen following an advice<br />

of Mr. Pierre ROCHUS, an engineer from Liège Space Center (CSL). For example, this<br />

glue was already used for the Planck satellite thermal test at CSL. The reasons that led<br />

us to choose the glue, are:<br />

25


• We do not want to drill or manufacture the CSK structure in any way. Indeed, any<br />

modication of this structure involves that new space qualication tests must be<br />

performed to validate it, which is not desired in our case.<br />

• Pumpkin has also developed solar panel clips to facilitate the integration of solar<br />

panels on the CSK structure [23]. However, in our case, these clips interfere with<br />

some solar cells (as shown in Figure 2.7) and then, they can not be used to attach<br />

our solar panels to the CSK structure.<br />

Figure 2.7: Interference between solar panel clips and <strong>OUFTI</strong>-1 solar cells [15]<br />

2.3.3 Antenna support<br />

<strong>OUFTI</strong>-1 uses two antennas, one for the downlink transmission in the Very High Frequency<br />

(VHF) band (145 MHz) and the other for the uplink transmission in the Ultra<br />

High Frequency (UHF) band (435 MHz). These antennas are made of a cupro-beryllium<br />

alloy.<br />

The antenna deployment mechanism is managed by the Mechanism (MECH) subsystem<br />

[24]. The antenna support is made of Al − 5754 and it is also hard-anodized, as the<br />

rails and feet of the skeleton, to ensure the electrical insulation of the antennas.<br />

The principle of the deployment mechanism is the following one: each antenna is rolled<br />

around the guide rails and is attached to it thanks to a wire of "Dyneema", which is<br />

an Ultra High Molecular Weight Polyethylene (UHMWPE) ber. Thirthy minutes after<br />

the deployment from the P-POD (which is a general requirement to avoid any collision<br />

between the CubeSats), the Electrical Power Supply (EPS) subsystem will deliver some<br />

electrical current to a thermal knife composed of a Titanium (T i) wire. When a current<br />

goes through this wire, it heats by Joule's eect: P = U I. This wire heats until it reaches<br />

the "Dyneema" melting point. At this moment, the antennas are released and deploy.<br />

26


This support is mounted on the face −X. This conguration is possible thanks to the<br />

restricted dimensions of the support (as shown in Figure 2.8).<br />

Figure 2.8: Conguration and dimensions of the antenna support<br />

Indeed, as already mentioned, the CubeSat standard requires that the access port area<br />

will not be obstructed. However, in our case, the only port which would be obstructed is<br />

the port used for the MHX transceiver (as shown in Figure 2.9). And, as it was already<br />

mentioned, <strong>OUFTI</strong>-1 does not use this function of the FM 430. So, there is no problem to<br />

put the antenna support at this place.<br />

Figure 2.9: Description of the port obstructed by the antenna support<br />

The xing method is the same as for the solar panels. The support will be glued on<br />

the skeleton using epoxy Stycast 2850 (with catalyst 24 LV).<br />

2.3.4 Permanent magnet and hysteretic bars<br />

To maintain the attitude of <strong>OUFTI</strong>-1, a control system is needed. This one is managed<br />

by the Attitude Determination and Control System (ADCS) subsystem [25].<br />

27


In our case, because of the limited available power and the fact that the payloads do<br />

not require a precise attitude control, it was decided to use a passive control system. This<br />

system comprises one permanent magnet, with dimensions of 20 × 4 × 4 mm, made of<br />

AlNiCo 5 (an Aluminum-Nickel-Cobalt alloy) and four hysteretic bars, with dimensions<br />

of 80 × 1 × 1 mm, made of P ermenorm 5000 H2. The principle is the following one: the<br />

permanent magnet tends to align its magnetization axis with the local Earth's magnetic<br />

eld (as shown in Figure 2.10). The hysteretic bars allow to limit, by hysteretic damping,<br />

the rotation rate of the satellite to an acceptable value (which will have to be determined<br />

by tests).<br />

Figure 2.10: Principle of the <strong>OUFTI</strong>-1 ADCS subsystem [25]<br />

Last year, a solution with only two hysteretic bars was implemented (the conguration<br />

is available in Figure 2.11).<br />

Figure 2.11: Last year's conguration of the ADCS<br />

28


However, in this conguration, a critical point that was not taken into account, is<br />

the saturation of the hysteretic bars by the permanent magnet. Indeed, this phenomenon<br />

limits the magnetic moment of the magnet, which presents a problem because this magnetic<br />

moment is essential to maintain the satellite attitude so, stronger it is, better is the solution.<br />

In accordance with these observations, the conguration with the greatest distance between<br />

the magnet and the hysteretic bars seems to be preferable. This situation is particularly<br />

tricky in nanosatellites because the available space is strongly restricted. Moreover, the<br />

principal axis of each hysteretic bar should be as close as possible of the plane perpendicular<br />

to the magnetization axis of the magnet. Any displacement from this plane leads to the<br />

apparition of an undesired component of the magnetic eld H magnet directed along the bar.<br />

This component results in a displacement of the working point on the hysteresis cycle, and<br />

it therefore aects the damping eciency of the hysteretic bar. An other problem is that<br />

the magnetic eld inside the hysteretic bars should be less than 30 A/m, which is a typical<br />

value of the Earth's magnetic eld on <strong>OUFTI</strong>-1 orbit. This suggests that the bars should<br />

ideally be placed as depicted in Figure 2.12.<br />

Figure 2.12: Ideal positioning of the hysteretic bars<br />

First conguration<br />

Firstly, it was decided to place the magnet on the battery support, close to the center<br />

of face −X. Its orientation was chosen in accordance with the requirement of the Radio<br />

Frequency (RF) subsystem: "The magnetization axis of the magnet must be parallel to<br />

the antennas' plane". The hysteretic bars were then placed as far as possible from the<br />

permanent magnet. Unfortunately, this conguration has showed an important inuence<br />

of the magnet on the hysteretic bars, which reduces the damping eciency. Figure 2.13<br />

shows the magnetic eld due to the magnet in the vicinity of the hysteretic materials.<br />

Therefore, a new conguration with a greater distance between hysteretic bars and the<br />

permanent magnet must be found.<br />

29


Figure 2.13: Magnetic eld obtained in the hysteretic bars for the rst conguration [25]<br />

Second conguration<br />

Then, it was decided to shift the magnet and to place it in the corner between faces −X,<br />

+Y and +Z. This new arrangement ensures the lowest possible inuence of the magnet<br />

on the hysteretic bars (as shown in Figure 2.14).<br />

Figure 2.14: Magnetic eld obtained in the hysteretic bars for the second conguration<br />

[25]<br />

Final conguration<br />

The critical point of the elongation of the hysteretic bars was not studied yet, and it<br />

appeared that it was crucial to have a high elongation. So, in order to satisfy this necessity,<br />

a new design was imagined. This one considers the possibility of using two parallel bars<br />

in each direction perpendicular to the magnet, for a total of four bars placed on corners<br />

between faces +X/ + Y , +X/ − Y , +Y/ − Z and −Y/ − Z respectively. So, to obtain<br />

the greatest distance between the magnet and the hysteretic bars, this one is placed in the<br />

middle of the corner between faces −X and +Y (as shown in Figure 2.15).<br />

30


Figure 2.15: Final conguration of the ADCS<br />

This improves the overall system eciency and leads to an optimal solution (the magnetic<br />

eld inside each hysteretic bar is represented in Figure 2.16).<br />

Figure 2.16: Magnetic eld obtained in the hysteretic bars for the nal conguration [25]<br />

Pictures of the permanent magnet and hysteretic bars (these ones are not yet manufactured<br />

to the good dimensions) are shown in Figure 2.17.<br />

31


Figure 2.17: Pictures of the ADCS components [25]<br />

2.3.5 Electronic cards and battery support<br />

<strong>OUFTI</strong>-1 disposes of ve electronic cards. These cards are all based on the P C 104<br />

standard [26]. The particularity of this standard is the P C 104 connector. This device,<br />

with 104 pins welded on it, allows a direct connection between all the cards of <strong>OUFTI</strong>-1.<br />

The PCBs are all made of Flame Resistant (FR) 4 that belongs to the class of Glass Fiber<br />

Reinforced Plastic (GFRP) material.<br />

The conguration for these cards, shown in Figure 2.18, is based on the following<br />

requirements:<br />

• The FM 430 card is placed in the bottom of the satellite, because it is on this card<br />

that the access ports are placed. So, these ports have to be placed in front of the<br />

holes of the chassis provided for this purpose, which are situated on the base plate.<br />

• The OBC 2 card is placed just above the FM 430 card. The reason is that these two<br />

cards have to communicate continuously so, to avoid a complicate cabling path, they<br />

are placed as close as possible one from the other. These two cards are managed by<br />

the student in charge of the OBC subsystem [27].<br />

• Then, we choose to integrate the EPS card, because it is on this card that there<br />

is most components and, consequently, it is the card with the greatest mass. So,<br />

its location aects seriously the position of the CoG and, to respect the CubeSat<br />

standard requirement on the CoG location, it is better to place it as close as possible<br />

from the CubeSat geometric center.<br />

• For the same reason, the battery support comes next. Further detail will be available<br />

on chapter 3.<br />

• The fourth card is the xEPS one. The reason is the same that one for the two OBC<br />

cards (these cards have to be linked by several wires). However, in addition, we<br />

32


Figure 2.18: Conguration of <strong>OUFTI</strong>-1 electronic cards<br />

33


consider the possibility of oshoring the protection circuit of the batteries on one<br />

electronic card and, in this case, the card with least components is the xEPS. The<br />

last two cards are managed by the student in charge of EPS and xEPS subsystems.<br />

• Finally, the Telecommunication (COM) card is placed on the top of the satellite, close<br />

to the end plate. This card is managed by students in charge of the COM subsystem<br />

([28] and [29]).<br />

Schematics and exact dimensions of each of these cards are available in appendix B,<br />

such as the available volume for each of them.<br />

Each of the components presented in this section, is separated from the others by spacers,<br />

made of Al − 6061 T 6, with variable dimensions (4 mm, 15 mm and 19 mm). The<br />

whole is xed to the main structure with the help of 4 M3 × 8 cm stainless steel endless<br />

screws, which are screwed on the base plate. On the other end of these screws, there are<br />

four midplane standos [30], made of Al −6061 T 6, that allow to connect the screws to the<br />

chassis. These midplane standos are also xed using M3 nuts. Schematics and pictures<br />

of these xations are available in appendix C.<br />

Pictures of the engineering models of each of these cards are available in Figure 2.19.<br />

Figure 2.19: Pictures of the engineering models of <strong>OUFTI</strong>-1 electronic cards<br />

34


2.4 Physical properties<br />

The physical properties (including CoG and inertia) are of primary importance for the<br />

denition of a 3-dimensional body. In addition, other students, like the ones in charge of<br />

ADCS [25] or MECH [24] subsystems, need them to perform their simulations.<br />

In this section, the Catia software will be used to determine theCoG location and<br />

inertia properties of <strong>OUFTI</strong>-1. When a complete engineering model of the satellite will be<br />

available, these values can be veried experimentally [31].<br />

2.4.1 Catia modeling<br />

Last year, a detailed model of the satellite was carried out using the Catia software [32].<br />

During this year, this model was rened to obtain the nal conguration of the satellite.<br />

This conguration is shown in Figure 2.20.<br />

This type of modeling, which is widely used in mechanical industries, presents several<br />

advantages:<br />

• Firstly, it allows to have an overall view of each component's location within the<br />

CubeSat. This is very helpful to verify that a given component could be placed at<br />

a specic location or to create an appropriate integration procedure for the satellite<br />

[33]. In addition, Catia les are compatible with the Samcef software [34], which is<br />

used to perform the FE analysis. So, the two can be combined to take the best of the<br />

two worlds (Catia for Computer-Aided Design (CAD) modeling and Samcef for FE<br />

analysis) and to facilitate the task of the engineer who is in charge of the structural<br />

subsystem.<br />

• Then, Catia can calculate several physical properties such as the location of the CoG<br />

(which is really helpful in our case considering the reference frame that we use) or<br />

the inertia properties. However, to be accurate, Catia requires the denition of each<br />

material used to manufacture the several components of the satellite.<br />

• Finally, Catia facilitates the creation of technical drawings.<br />

2.4.2 Center of gravity<br />

An other requirement imposed by the CubeSat standard is: "The CubeSat center of<br />

gravity shall be located within a sphere of 2 cm from its geometric center ". This requirement<br />

allows to ensure that each P-POD CoG remains located inside a specic volume.<br />

35


Figure 2.20: Exploded view of <strong>OUFTI</strong>-1<br />

36


In our case, this requirement can be easily veried considering our particular reference<br />

frame. The results obtained for <strong>OUFTI</strong>-1 in its nal conguration are listed in Table 2.1.<br />

Axes Coordinates (in mm)<br />

X -0.139<br />

Y -2.558<br />

Z 0.691<br />

Table 2.1: Coordinates of the <strong>OUFTI</strong>-1 CoG<br />

So, the distance between the geometric center of <strong>OUFTI</strong>-1 and its CoG can be calculated<br />

using Equation (2.4.1).<br />

√<br />

d = x 2 CoG + y2 CoG + z2 CoG<br />

= 2.6533 mm (2.4.1)<br />

Therefore, the requirement is respected.<br />

2.4.3 Inertia<br />

For a continuous rigid body, the inertia tensor is given by Equation (2.4.2).<br />

⎛<br />

⎞<br />

I xx I xy I xz<br />

I = ⎝I yx I yy I yz<br />

⎠ (2.4.2)<br />

I zx I zy I zz<br />

where:<br />

I xx =<br />

∫ (y 2 + z 2) dm (2.4.3)<br />

I yy =<br />

∫ (x 2 + z 2) dm (2.4.4)<br />

I zz =<br />

∫ (x 2 + y 2) dm (2.4.5)<br />

∫<br />

I xy = I yx = −<br />

∫<br />

I xz = I zx = −<br />

∫<br />

I yz = I zy = −<br />

(x y) dm (2.4.6)<br />

(x z) dm (2.4.7)<br />

(y z) dm (2.4.8)<br />

37


I xx , I yy and I zz are called the moments of inertia while the other elements are called<br />

the products of inertia.<br />

This tensor can be calculated from any point of the structure. However, to facilitate<br />

its calculation, it is preferable to dene it from the CoG. Thus, the inertia tensor linked to<br />

our reference frame is given in Equation (2.4.9).<br />

⎛<br />

2.104 6.025 10 −2 1.69 10 −2 ⎞<br />

I CoG = ⎝6.025 10 −2 1.757 −9.641 10 −2 ⎠ × 10 6 g mm 2 (2.4.9)<br />

1.69 10 −2 −9.641 10 −2 2.019<br />

In addition, the principal moments of inertia can be calculated, as well as the principal<br />

axes, to completely dene the inertia properties of <strong>OUFTI</strong>-1. They are given in Equations<br />

(2.4.10) to (2.4.12) and (2.4.13) to (2.4.15) respectively.<br />

M 1 = 1.715 × 10 6 g mm 2 (2.4.10)<br />

M 2 = 2.05 × 10 6 g mm 2 (2.4.11)<br />

2.5 Mass budget<br />

M 3 = 2.114 × 10 6 g mm 2 (2.4.12)<br />

⎛ ⎞<br />

−0.159<br />

A 1 = ⎝ 0.938 ⎠ (2.4.13)<br />

0.307<br />

⎛ ⎞<br />

−0.042<br />

A 2 = ⎝ 0.305 ⎠ (2.4.14)<br />

−0.952<br />

⎛ ⎞<br />

0.986<br />

A 3 = ⎝0.164⎠ (2.4.15)<br />

0.009<br />

As already mentioned on chapter 1, one of the general constraints for a CubeSat is:<br />

"Each single CubeSat shall not exceed 1 kg mass ".<br />

To ensure that <strong>OUFTI</strong>-1 respects this requirement, each component of the CubeSat<br />

was weighted. However, at the end of this year, some of them are not yet nished. For<br />

38


these ones, an estimation of the mass was presented in a "best case/worst case" philosophy.<br />

The mass budget nally obtained is presented in Table 2.2. It is rstly divided between<br />

the several subsystems in order to focus on the weight of each component to identify the<br />

main mass consuming ones. Then, a general table is presented and the total mass of<br />

<strong>OUFTI</strong>-1 is calculated. It should be noted that each mass was rounded to upper 0.01 g,<br />

which brings some safety to the measurements. However, to obtain a reliable nal result,<br />

a Safety Factor (SF) of 2% was applied to the total mass of <strong>OUFTI</strong>-1.<br />

2.6 Summary<br />

Through this chapter, the general conguration of <strong>OUFTI</strong>-1 was discussed.<br />

First, a reference frame was dened to allow to locate precisely each component of the<br />

CubeSat. The advantages of the selected frame were also highlighted.<br />

Then, the organization of the several subsystems was exposed. For each of them, a<br />

detailed description of the location, xing methods and general physical and mechanical<br />

properties was presented. These descriptions allow to bring an overall view of the satellite<br />

to the profane.<br />

Finally, the physical properties of <strong>OUFTI</strong>-1 were determined, and their fulllments of<br />

the ESA requirements were veried.<br />

39


Subsystem: Structure & Conguration<br />

Parts Components Number Best case (g) Worst case (g)<br />

Base plate 1 35.72<br />

Foot 1 1.32<br />

Grower washer M3 1 0.06<br />

Screw M3 × 4 mm 1 0.37<br />

Drilled foot 3 3.63<br />

Base plate<br />

Spring plunger 2 1.2<br />

Nut M3 2 1.1<br />

Plastic stem 1 0.15<br />

Deployment switch 1 2.44<br />

Screw M1 × 1 cm 2 0.44<br />

Nut M1 1 0.18<br />

Shym 4 0.62<br />

Subtotal 47.23<br />

End plate 1 26.29<br />

Foot 4 5.27<br />

End plate Shym 4 0.62<br />

Grower washer M3 4 0.21<br />

Screw M3 × 4 mm 4 1.48<br />

Subtotal 33.87<br />

Chassis<br />

Chassis 1 63.39<br />

Screw M3 × 4 mm 10 3.19<br />

Subtotal 66.58<br />

Base panel 1 38.5<br />

Solar cell 2<br />

Base panel Kapton lm /<br />

Adhesive /<br />

10.5<br />

Coverglass /<br />

Subtotal 49<br />

End panel 1 33<br />

Solar cell 2<br />

End panel Kapton lm /<br />

Adhesive /<br />

8.9<br />

Coverglass /<br />

Subtotal 41.9<br />

40


Parts Components Number Best case (g) Worst case (g)<br />

Chassis's panel 3 94<br />

Solar cell 6<br />

Chassis's panels Kapton lm /<br />

Adhesive /<br />

31.3<br />

Coverglass /<br />

Subtotal 125.3<br />

Box 1 27.33<br />

Battery support<br />

Cover 1 19.24<br />

Screw M3 × 8 mm 4 1.86<br />

Subtotal 48.43<br />

Endless screw 4 14.6<br />

Spacer 4 mm 4 0.31<br />

Spacer 15 mm 12 3.39<br />

Fixations<br />

Spacer 19 mm 4 1.43<br />

Midplane stando 4 3.03<br />

Washer 4 0.38<br />

Screw M3 × 4 mm 4 1.28<br />

Nut M3 4 1.31<br />

Subtotal 25.73<br />

Total 438.04<br />

Subsystem: Mechanism<br />

Parts Components Number Best case (g) Worst case (g)<br />

Antennas<br />

435 MHz 1 1.25<br />

145 MHz 1 3.46<br />

Subtotal 4.71<br />

Antenna support<br />

Support 1 26<br />

Deployment mechanism 1 1<br />

Subtotal 27<br />

Thermal knife Thermal knife 1 0.5<br />

Subtotal 0.5<br />

Total 32.21<br />

Subsystem: Attitude & Determination Control System<br />

Parts Components Number Best case (g) Worst case (g)<br />

Permanent magnet Permanent magnet 1 2 3<br />

Subtotal 2 3<br />

Hysteretic bars Hysteretic bar 4 2 3<br />

Subtotal 2 3<br />

Total 4 6<br />

41


Subsystem: Thermal Control<br />

Parts Components Number Best case (g) Worst case (g)<br />

General Heat sensor 3 3 6<br />

Subtotal 3 6<br />

Heater 2 2 3<br />

Batteries<br />

Thermostat 4 0.75 0.85<br />

Wire 8 8 10<br />

Thermal insulation 1 0.02 0.08<br />

Subtotal 10.77 13.93<br />

Cu strap 1 5 10<br />

EPS<br />

Bolt M3 3 9 18<br />

Resistance 2 2 3<br />

Subtotal 16 31<br />

Total 29.77 50.93<br />

Subsystem: Electrical Power Supply<br />

Parts Components Number Best case (g) Worst case (g)<br />

EPS Card 1 64.67<br />

Subtotal 64.67<br />

xEPS Card 1 58.7<br />

Subtotal 58.7<br />

Battery Kokam<br />

Batteries SLB 603870H<br />

2 62 66<br />

Wire 4 4 8<br />

Subtotal 66 74<br />

Total 189.37 197.37<br />

Subsystem: On-Board Computer<br />

Parts Components Number Best case (g) Worst case (g)<br />

FM 430 Card 1 66.85<br />

Subtotal 66.85<br />

OBC 2 Card 1 50.47<br />

Subtotal 50.47<br />

Total 117.32<br />

Subsystem: Telecommunication<br />

Parts Components Number Best case (g) Worst case (g)<br />

COM Card 1 50 65<br />

Total 50 65<br />

42


<strong>OUFTI</strong>-1<br />

Subsystems Best case (g) Worst case (g)<br />

Structure & Conguration 438.04<br />

Mechanism 32.21<br />

Attitude & Determination Control System 4 6<br />

Thermal Control 29.77 50.93<br />

Electrical Power Supply 189.37 197.37<br />

On-Board Computer 117.32<br />

Telecommunication 50 65<br />

Adhesive 20 40<br />

Cabling 15 30<br />

Subtotal 895.71 976.87<br />

Safety coecient 1.02 1.02<br />

<strong>OUFTI</strong>-1 mass 913.63 996.41<br />

Table 2.2: Mass budget of <strong>OUFTI</strong>-1<br />

43


Chapter 3<br />

Design of a new support for the<br />

batteries<br />

3.1 Introduction<br />

Through this chapter, the design of a new support for the batteries will be studied.<br />

Firstly, the reasons that led to change the support created last year, will be exposed.<br />

Then, the entire development phase, from the initial idea to the manufacturing of the<br />

new support, will be discussed step by step. This phase was realized in accordance with<br />

the requirements of EPS and Thermal Control (THER) subsystems.<br />

3.2 Problems encountered with the last year's design<br />

The design that was imagined last year [15], is presented in Figure 3.1.<br />

Figure 3.1: Last year's design of the battery assembly<br />

44


In this design, the battery Printed Circuit Board (PCB) was suspended to the xEPS<br />

card, using four T i screws. The two PCBs are separated by four T i spacers and thermal<br />

washers are placed at the level of the battery PCB to isolate it thermally.<br />

The main problem of this design was the available space between the batteries and the<br />

electronic cards situated under and above them, as it is shown in Figure 3.2.<br />

Figure 3.2: Available space under and above the batteries<br />

The available space between the lower battery and the highest component above the<br />

EPS card is given in Equation (3.2.1) and the distance between the upper battery and the<br />

highest component under the xEPS card is given in Equation (3.2.2).<br />

d Battery - EPS = 5.57 − 4.8 = 0.77 mm (3.2.1)<br />

d Battery - xEPS = 4.83 − 1.5 = 3.33 mm (3.2.2)<br />

So, it can already be noted that the position of the batteries inside the satellite is not<br />

optimal. In addition, due to the restricted value of the distances which separate these<br />

components, several problems can occured, especially concerning the vibrations. Indeed,<br />

the batteries could collide one of the two electronic cards and cause important damages on<br />

it.<br />

In addition to this restricted available space, it appeared, after a test under vacuum<br />

conditions, that the batteries bulged (as shown in Figure 3.3). This increase the problem of<br />

the available space and, moreover, can aect the electrical performance of the batteries. To<br />

avoid this problem, the solution which was investigated, was to encapsulate the batteries<br />

inside a box.<br />

45


Figure 3.3: Batteries during and after the vacuum test [15]<br />

However, the problem for the design of this box is double:<br />

• Firstly, the restricted available space (2.05 mm on each side in the optimal conguration)<br />

does not allow to design a reliable box. Indeed, the vibrations generated<br />

during the launch phase might reach amplitudes that rise above this value. So, the<br />

box might damage the electronic cards situated close to it.<br />

• Then, the mass budget of the satellite in the last year's nal conguration is already<br />

higher than the 1 kg mass allowed by the CubeSat standard (the total mass of<br />

<strong>OUFTI</strong>-1 in this conguration was 1.032 kg).<br />

Thus, we had to give up this solution and to create a new design, including a box, to<br />

encapsulate the batteries.<br />

3.3 Available mass budget<br />

The rst step to realize, when a new design is created, is to calculate the available<br />

mass budget. To obtain this budget, we have to substract the mass of the components<br />

that composed the old design from the total mass of <strong>OUFTI</strong>-1 obtained for the last year's<br />

conguration. Then, the result must be substracted from the maximum acceptable mass<br />

of 1 kg to obtain the available mass for the new support. This calculation is performed in<br />

Table 3.1 and Equation 3.3.1.<br />

m available = m max, acceptable<br />

−<br />

SF<br />

= 1000 ( 1031.4148<br />

1.02 − 1.02<br />

= 103.79 g<br />

( mtot, last year<br />

− m last year<br />

SF<br />

′ s design<br />

)<br />

− 134.58<br />

)<br />

(3.3.1)<br />

46


Components<br />

Mass (g)<br />

4 spacers 25 mm 4 × 0.475 = 1.9<br />

4 T i screws 4 × 0.86 = 3.44<br />

4 T i spacers 4 × 0.29 = 1.16<br />

4 T i nuts 4 × 2.32 = 9.28<br />

2 heaters 2 × 1.5 = 3<br />

Heaters' control system 10<br />

Radiative insulation 15<br />

16 thermal washers 16 × 0.08125 = 1.3<br />

Battery PCB 15.5<br />

2 batteries Kokam SLB 603870H 2 × 33 = 66<br />

4 wires 4 × 2 = 8<br />

Total 134.58<br />

Table 3.1: Available mass budget<br />

To this result, some important corrections must be added. These corrections, which<br />

came from revisions or improvements of designs, are related in Table 3.2.<br />

Components Last year's mass (g) This year's mass (g) Dierences (g)<br />

Thermal knife 19 0.5 + 18.5<br />

Permanent magnet 13 3 + 10<br />

Hysteretic bars 14 3 + 11<br />

COM card 75 65 + 10<br />

Total 121 71.5 + 49.5<br />

Table 3.2: Important corrections in the mass budget<br />

So, the total available mass budget for the new support is: 103.79 + 49.5 = 153.29 g.<br />

3.4 Initial idea<br />

The base idea of the new support was imagined by Jean-Philippe NOEL, who is in<br />

charge of the THER subsystem [35]. The idea is the following one: instead of placing the<br />

batteries one on the other, it was a better idea to place them one next to the other. This<br />

new conguration allows to release more space between the batteries and the electronic<br />

cards for integrating the box. An illustration of this new design is available in Figure 3.4.<br />

47


The concept is the following one:<br />

Figure 3.4: Initial idea for the new design<br />

• The support is composed of two dierent parts: one box, in which the batteries will<br />

be placed, and one cover, which will prevent the batteries' bulge. These two parts<br />

will be connected together by several screws.<br />

• The assembly will be supported by the four endless screws, just like all the electronic<br />

cards. So, new spacers will have to be designed to replace the 25 mm ones that previously<br />

separated the EPS card from the xEPS card. Their length will be determined<br />

later.<br />

• The box is composed of a "base plate" with reinforcers. The holes situated at the<br />

level of the batteries' width are there to allow the wires to come out of the box.<br />

Starting from this idea, the support was adapted to the <strong>OUFTI</strong>-1 dimensions. It was<br />

decided to place it with the box on the side of the xEPS card. This detail has to be noted<br />

because the four endless screws are not placed symmetrically inside the CubeSat so, the<br />

direction in which the support will be integrated, takes on all its importance. This choice<br />

is based on the CubeSat standard requirement about the CoG location already mentioned<br />

in section 2.4.2. The result obtained is given in Figure 3.5.<br />

48


Figure 3.5: Adaptation of the base idea to the <strong>OUFTI</strong>-1 dimensions<br />

3.5 Battery selection<br />

Once the four holes for the endless screws are drilled, the available space for the batteries<br />

is restricted to the dimensions shown in Figure 3.6.<br />

Figure 3.6: Available space for including the batteries<br />

With these dimensions in mind, the battery selection can start.<br />

Last year, it was decided to use Lithium (Li) batteries for their compact form and their<br />

high specic capacity. Two dierent types of batteries were considered: Varta and Kokam.<br />

49


A test under vacuum conditions was performed on these batteries and it appeared that<br />

the Kokam bulged and not the Varta. However, it should be noted that batteries from<br />

Varta presented several uncertainties in relation to their electrical capacity evolution in<br />

time. In addition, these batteries were no longer provided by the industry and, for the<br />

remaining ones, the storage conditions were not known. So, it was decided to use batteries<br />

from Kokam.<br />

The main selection criteria are: dimensions, mass, electrical capacity and, last but<br />

not least, reliability. Accounting for these restrictions, three batteries were selected. Their<br />

properties are available in Table 3.3. It should be noted that the electrical capacity of these<br />

batteries is given for standard conditions (T = 25 ◦ C, V = 2.7 − 4.2 V and q = 0.5 C).<br />

Batteries<br />

Length Width Thickness Mass Electrical capacity<br />

(mm) (mm) (mm) (g) (Ah)<br />

SLP B 554374H 73.5 ± 0.5 42.5 ± 0.5 5.4 ± 0.2 32 ± 1 1.25<br />

SLB 603870H 69.5 ± 0.5 37.5 ± 0.5 6.5 ± 0.2 32 ± 1 1.5<br />

SLP B 723870H4 70 ± 0.5 37.5 ± 0.5 7.2 ± 0.2 37 ± 1.5 1.5<br />

Table 3.3: Properties of several batteries from Kokam [36]<br />

Considering the electrical capacity, <strong>OUFTI</strong>-1 needs 1.2 Ah. So, the rst battery can not<br />

be selected because the Margin of Safety (MoS) is too small in this case. In addition, this<br />

battery has dimensions that do not allow to include it in the available space described in<br />

Figure 3.6. Now, if the physical properties are also taken into account, the better solution<br />

is to use the battery Kokam SLB 603870H, which is lighter and smaller than the other<br />

one. A picture of this battery is presented in Figure 3.7. Its datasheet is available in<br />

appendix D.<br />

Figure 3.7: Battery Kokam SLB 603870H<br />

This choice, which fullls all the requirements of Structure & Conguration (STRU)<br />

50


subsystem, was validated by the EPS team. So, the mass and dimensions of the batteries<br />

are now known and a rst guess of how arrange the reinforcers can be realized.<br />

3.6 Material selection<br />

An other important step for realizing the design of the support, is the selection of the<br />

material that will be used to manufacture it. Indeed, the thickness of the dierent parts<br />

that must prevent the batteries' bulge, was determined in relation to the properties of the<br />

selected material.<br />

3.6.1 Denition of the requirements<br />

Firstly, the requirements imposed to the material must be dened precisely.<br />

Structural requirements<br />

The rst feature of the battery support is its structural integrity during the launch.<br />

Indeed, it will have to support the harsh vibration environment without causing damages<br />

to the other components of the satellite. So, the strength is an important property that<br />

will ensure the structural integrity of the support. The toughness is also an important<br />

property to sustain the stage ignition/separation, fairing jettisoning and POGO eects<br />

during launch. Then, the structural requirements can be expressed as:<br />

• The material shall be rigid (high Young's modulus).<br />

• The material shall have a good resistance in traction/compression.<br />

• The material shall have a toughness of at least 25 MP a m 0.5 (which is a value typically<br />

used in the space applications).<br />

Thermal requirements<br />

Firstly, the material shall resist to the operating temperature range of the batteries<br />

([0 ◦ C − 40 ◦ C]). Then, thermal variations inside the CubeSat can disturb the batteries'<br />

performances. These variations can be limited by using a material with a high thermal<br />

inertia. In addition, the heat produced by the heaters must only serve to maintain the<br />

batteries' temperature in an acceptable range. Any waste of energy could cause important<br />

problems on the power distribution inside the satellite. So, a material with a low thermal<br />

conductivity is preferable to keep the heat inside the support. Finally, thermal cycling can<br />

also induce deformations of the support. However, in this case, the support could not play<br />

its role of preventing the batteries' bulge. This eect can be limited if the Coecient of<br />

51


Thermal Expansion (CTE) of the material is not too high (e.g, CTE of the CSK structure<br />

is: 20 − 25 µstrain/K). Then, the thermal requirements can be expressed as:<br />

• The material shall resist to temperature range [−10 ◦ C − 50 ◦ C] (by taking a security<br />

margin of 10 ◦ C).<br />

• The material shall shield batteries against thermal variations.<br />

• The material shall have a low thermal conductivity (however, if this requirement is<br />

not respected, it is possible to use some thermal insulators between the batteries and<br />

their support).<br />

• The material shall have a low CTE.<br />

Outgassing requirements<br />

These requirements are imposed by the ESA [37]:<br />

• Total Mass Loss (TML) of the material shall be ≤ 1%.<br />

• Collected Volatile Condensable Material (CVCM) of the material shall be ≤ 0.1%.<br />

Presetting<br />

In accordance with all these requirements, four materials were selected: Al alloys,<br />

Carbon Fiber Reinforced Plastic (CFRP), GFRP and T i alloys. All of these materials<br />

fulll all the requirements and they are all typically used in space applications. Their<br />

general properties are available in Table 3.4.<br />

Materials Densities (kg/m 3 ) Young's moduli (GP a) Yield stresses (MP a)<br />

Al alloys 2500 − 2900 68 − 80 95 − 610<br />

CFRP 1500 − 1600 69 − 150 550 − 1050<br />

GFRP 1750 − 1970 15 − 28 110 − 192<br />

T i alloys 4400 − 4800 110 − 120 750 − 1200<br />

Table 3.4: Materials properties<br />

52


3.6.2 Objective function and CES software<br />

Selecting particular materials and manufacturing process is a complex task that needs<br />

several iterations to be carried through. The procedure used here is based on references<br />

[38] and [39].<br />

The rst step consists in dening one or several objective functions that allow to classify<br />

the materials. In our case, the objective is to obtain a reliable support with a minimal<br />

weight. So, two objective functions can be dened (see Equations 3.6.1 and 3.6.2).<br />

M 1 = E1/2<br />

ρ<br />

(3.6.1)<br />

M 2 = σ2/3 y<br />

(3.6.2)<br />

ρ<br />

where M 1 and M 2 are the performance index to maximize, E is the Young's modulus of<br />

the material, ρ is its density and σ y is its yield stress.<br />

According to these functions, it is possible to use the CES software to realize the material<br />

selection. This powerful software, developed by the Prof. M. ASHBY, includes a<br />

complete material database coupled with a program that classies material with regard to<br />

predined properties.<br />

In our case, the four materials listed in section 3.6.1 were classied using the CES<br />

software. The graphs obtained are presented in Figures 3.8 and 3.9.<br />

Figure 3.8: Classication by density and Young's modulus<br />

53


Figure 3.9: Classication by density and yield stress<br />

So, the nal classication of these materials is:<br />

1. CFRP<br />

2. Al alloys<br />

3. GFRP<br />

4. T i alloys<br />

However, an other important criterion to take into account is the cost. Indeed, each of<br />

these materials has a given cost, which is not the same for each of them. Moreover, the<br />

cost is also closely linked to the manufacturing process which can be used to modeled the<br />

material into the desired form.<br />

For this reason, Al alloys become more attractive than CFRP. In addition, the choice<br />

of Al alloys allows to create a support with more restricted dimensions, which is really<br />

important considering the available space between EPS and xEPS cards.<br />

The selection of a particular alloy was then based on an ESA requirement which stipulates<br />

that: "Aluminum 7075 or 6061 shall be used for both the main CubeSat structure<br />

and the rails. If other materials are used the developer shall submit a Deviation Waiver<br />

Approval Request (DAR) and adhere to the waiver process ".<br />

In this requirement, the denition for a "CubeSat main structure" is the following one:<br />

a main structure is a structure which supports, in addition to its own weight, the weight<br />

54


of other signicant components of the satellite, which is precisely the case of the battery<br />

support.<br />

For availability reasons, the Al − 7075 T 6 alloy was selected [40].<br />

3.7 Thermal control's devices selection<br />

Now, the selection of the dierent devices that perform the thermal control of the<br />

batteries, will be discussed. It should be noted that this selection was realized by the<br />

THER team. Further detail on each of these devices can then be found in reference [35].<br />

In this section, we will only draw a list of these components and, for each of them, the<br />

advantages that led to this particular choice.<br />

3.7.1 Heaters<br />

To maintain the batteries' temperature range in acceptable values, the use of heaters is<br />

necessary. In our case, the heaters are polyimide thermofoil heaters from Minco industry<br />

(illustrated in Figure 3.10) [41].<br />

Figure 3.10: Polyimide thermofoil heater<br />

These heaters present several advantages:<br />

• They are available in a lot of dierent forms and dimensions. In our case, we chose<br />

heaters that had the dimensions closest to the batteries' ones.<br />

• They are lightweight.<br />

• They can be glued directly on the batteries. This particularity is widely used in space<br />

applications.<br />

• They are space-qualied.<br />

55


The last property to determine is the value of the resistance. Thermal simulations<br />

performed last year [12] allow to determine that the power necessary to heat the batteries, is<br />

250 mW . Assuming that heaters are supplied by a tension of 2.5 V (which is a conservative<br />

approach because the tension of the batteries can not pass under 2.7 V ), the resistance<br />

can be easily calculated (see Equation 3.7.1).<br />

P = U I = U 2<br />

R ⇒ R = U 2<br />

P = 2.52 = 25 Ω (3.7.1)<br />

0.25<br />

The available heater that is closest to this value is a heater with a resistance of 23.7 Ω,<br />

which is again a conservative approach because the power generated will be higher than<br />

250 mW .<br />

The heaters will be glued, as already mentioned, directly on the batteries, on their face<br />

+Z (which corresponds to the side of the box).<br />

3.7.2 Heaters' control system<br />

Two solutions were investigated to realize the heaters' control system:<br />

• Creation of an electrical circuit that used heat sensors to decide when the heaters<br />

have to be supplied.<br />

• A thermostat that is self-ecient.<br />

In accordance with the KISS philosophy, the second solution was preferred because it<br />

is more reliable.<br />

The selected thermostats are Klixon 4BT − 2 [42], illustrated in Figure 3.11 (the dimensions<br />

are given in inches).<br />

Figure 3.11: Thermostat Klixon 4BT − 2<br />

56


These thermostats were chosen because:<br />

• They have very small dimensions (diameter = 6.68 mm and thickness = 2.04 mm).<br />

• They are lightweight (mass = 0.2 g).<br />

• They are space-qualied (conform to qualication MIL-S-24236/13).<br />

In our case, the selected thermostat is an open-on-rise one. The mechanism of these<br />

thermostats is the following one:<br />

• When the temperature of the batteries passes under 7.2 ◦ C, the thermostats supply<br />

the heaters. This particular value was chosen by the THER subsystem to avoid that<br />

the temperature of the batteries becomes negative. Indeed, the tolerance guaranteed<br />

by the manufacturer is 4.4 ◦ C, and the rst available value above this one is<br />

7.2 ◦ C (4.4 ◦ C is also available but, in this case, the non-negativity of the batteries'<br />

temperature could not be assured at each time).<br />

• They continue until the temperature of the batteries pass above 23.9 ◦ C (± 4.4 ◦ C).<br />

At this moment, the thermostats open the circuit and the heaters are no longer<br />

supplied.<br />

Now, the problem is to determine how many thermostats are necessary to obtain a<br />

reliable control system. Dierent solutions were investigated:<br />

• The rst idea was to use four thermostats per battery. These ones will be connected<br />

2 by 2 in series, and the two parts will be connected in parallel. This solution is<br />

illustrated in Figure 3.12 (where the B stands for Battery, H for Heater and T for<br />

Thermostat).<br />

Figure 3.12: Connection with four thermostats<br />

This solution had to be given up because it is too expensive for this particular application.<br />

57


• So, it was decided to use two thermostats per battery (to prevent a non-uniform<br />

distribution from the temperature inside the battery). These ones will be placed on<br />

the face −Z of the batteries (which corresponds to the side of the cover). So, a notch<br />

must be manufactured on the cover. Now, it must be decided how these thermostats<br />

will be connected together. The Table 3.5 summarizes the dierent types of available<br />

congurations and problems that can occured.<br />

Thermostat breaks open Thermostat breaks closed<br />

Connection Heater will never be supplied so, The second thermostat will work<br />

in series the battery will freeze as if it was alone<br />

Connection The second thermostat will work Heater will be continuously supplied so,<br />

in parallel as if it was alone the battery will burn out<br />

Table 3.5: Dierent available congurations and problems for thermostats<br />

In accordance with these observations, it was decided to connect them in series.<br />

Indeed, for the connection in parallel, the second problem causes a constant supply<br />

of the heater. In this case, all the power available inside the satellite would be used<br />

to supply this heater, and the entire satellite would be lost. With the connection in<br />

series, in the worst case, one battery freezes but the rest of the satellite is safe.<br />

3.7.3 Thermal insulator<br />

Aluminum, which is the material chosen to manufacture the battery support, is a thermal<br />

conductor. So, to avoid all waste of energy during the batteries' heating, a thermal<br />

insulator must be placed between the batteries and their support.<br />

The selected insulator is a polyester netting, illustrated in Figure 3.13.<br />

Figure 3.13: Thermal insulator: Polyester netting<br />

This type of insulator is typically used as a spacer material to minimize conductive<br />

heat transfer between Multilayer Insulation (MLI) blanket layers. The netting material<br />

is chosen for its low outgassing characteristics and is specially cleaned to assure that it is<br />

residue free [43].<br />

One layer of insulator will be placed on each side of the batteries (the layer on the cover<br />

side will surround the thermostats).<br />

58


3.8 Final design of the support<br />

Now that all the components to include inside the support are dened and the material<br />

used to manufacture it is selected, the nal design of this one must be initiated.<br />

The dierent steps of the box's design are summarized here:<br />

• Firstly, the thickness of the box's "base plate" was xed to 1 mm. This restricted<br />

thickness can be obtained thanks to the selected material (Al − 7075 T 6). Indeed,<br />

this material allows to create proles with very low thickness (that is not the case<br />

of the CFRP for which the manufacturing process is much more complicated). This<br />

thickness of 1 mm was applied everywhere, except around the xations with the<br />

endless screws because it was these parts of the box that sustain the entire weight of<br />

the support. So, to avoid stress concentration that would lead to a fracture of the<br />

support, the thickness of these parts was increased to 2 mm. This step is illustrated<br />

in Figure 3.14.<br />

Figure 3.14: Thickness of the box's "base plate"<br />

• Then, a hole must be added on the side of face −Y to allow the P C 104 connector<br />

from the EPS card to pass next to the support and to reach the connector from the<br />

xEPS card. A set must also be taken into account to avoid any collision between the<br />

support and the connector due to vibrations. This set has a value of 0.5 mm on each<br />

side in our case. This step is illustrated in Figure 3.15.<br />

Figure 3.15: P C 104 connector's hole<br />

• The next step concerns the reinforcers. Their thickness was also xed to 1 mm. The<br />

reinforcers were designed at the upper dimensions of the batteries (in order to be<br />

able to include them inside the box in every case), except for the thickness, for which<br />

they were designed at their lower dimensions. Indeed, the main goal of this support<br />

59


is to prevent the batteries' bulge. So, this one has to constrain the batteries in every<br />

case to prevent this bulge. However, this particular thickness does not allow to x<br />

the screws that realize the connection between the box and the cover of the support.<br />

For this reason, the thickness of the reinforcement was increased to 6 mm around<br />

the connection holes. This step is illustrated in Figure 3.16.<br />

Figure 3.16: Reinforcers and connection holes<br />

• Finally, useless material was removed from several places (the two side of the support<br />

and the side reinforcers) to lighten the box. Some weld toes were also added to obtain<br />

the nal design of the box presented in Figure 3.17. It should be noted that the weld<br />

toes situated at the level of the two parts around the hole for the P C 104 connector,<br />

are bigger than the others. This choice was made to avoid any stress concentration<br />

inside the support's legs, which would lead to the fracture of this one.<br />

Figure 3.17: Final design of the box<br />

This nal design leads to a volume of 9.76 × 10 −6 m 3 , which corresponds to a mass of<br />

27.33 g. Schematics with exact dimensions are presented in Figure 3.18.<br />

60


Figure 3.18: Schematics of the box<br />

The dierent steps of the cover's design are summarized here:<br />

• Firstly, just as for the "base plate" and the reinforcers of the box, the thickness of<br />

the cover was xed to 1 mm. To avoid a waste of mass, the cover's dimensions were<br />

restricted to their minimal value. However, some reinforcements have to be added<br />

around the holes of the connection screws to avoid any stress concentration. These<br />

reinforcements allow to obtain a lighter design. This step is illustrated in Figure 3.19.<br />

Figure 3.19: Reinforcements around the connection holes<br />

61


• Then, the thermostats, that will be placed on the cover side, have to be in direct<br />

contact with the batteries. However, the role of the support, which is to prevent the<br />

batteries' bulge, must not to be forgotten. This fact does not allow to increase the<br />

height of the entire cover. So, some notches have to be manufacture in the cover to<br />

allow the thermostats' integration. This step is illustrated in Figure 3.20.<br />

Figure 3.20: Notches for the thermostats<br />

• An important fact that must be noted is that the thickness of the batteries is not<br />

constant. Indeed, there is a small part, including an electrical insulator, which is of<br />

negligible thickness. So, to avoid any bulge of the thicker part, some end stops must<br />

be added under the cover. This step is illustrated in Figure 3.21.<br />

Figure 3.21: End stops<br />

• Finally, as for the box, some weld toes are added to obtain the nal design of the<br />

cover presented in Figure 3.22.<br />

62


Figure 3.22: Final design of the cover<br />

This nal design leads to a volume of 6.869 × 10 −6 m 3 , which corresponds to a mass of<br />

19.24 g. Schematics with exact dimensions are presented in Figure 3.23.<br />

3.9 Validation of the design<br />

Now, the design which was nally obtained, must be validated.<br />

3.9.1 Mass budget<br />

Firstly, the necessary mass budget must be lower than the available mass budget estimated<br />

in section 3.3. This one is calculated in Table 3.6.<br />

63


Figure 3.23: Schematics of the cover<br />

64


Components<br />

Mass (g)<br />

Box 27.33<br />

Cover 19.24<br />

4 screws M3 × 8 mm 4 × 0.465 = 1.86<br />

4 spacers 4 mm 4 × 0.0775 = 0.31<br />

4 spacers 19 mm 4 × 0.3575 = 1.43<br />

2 batteries KOKAM SLB 603870H 2 × 33 = 66<br />

4 wires 4 × 2 = 8<br />

2 heaters 2 × 1.5 = 3<br />

4 thermostats 4 × 0.2125 = 0.85<br />

8 wires 8 × 1.25 = 10<br />

Thermal insulator 0.08<br />

Total 138.1<br />

Table 3.6: Necessary mass budget<br />

This budget is well lower than the available mass budget, which is of 153.29 g. So, our<br />

design is valid with regard to the mass.<br />

3.9.2 Static loads<br />

Then, the structural integrity of the support under static loads at which it will be<br />

submitted during the launch phase, must be veried.<br />

For this, some FE analysis have to be performed. To limit the modeling eorts, only<br />

the relevant structural parts will be considered. The modeling strategy for each part of<br />

the support is:<br />

• The box and the cover of the support are exible volumes made of Al − 7075 T 6.<br />

The properties of this material are:<br />

Young's modulus: E = 72.5 GP a<br />

Poisson ratio: ν = 0.33<br />

Density: ρ = 2800 kg/m 3<br />

• The batteries, which are also exible volumes, weight 32 g and have a volume of<br />

1.482 × 10 −5 m 3 in nominal conditions, which lead to a density of 2159.25 kg/m 3 .<br />

However, their Young's modulus can not be estimated and is not provided in the<br />

datasheets. To determine it, some static bench tests would be necessary but, in our<br />

case, these components will be represented by increasing the density of the support<br />

inside the area normally occuped by the batteries.<br />

65


• Other components, including thermostats, heaters and thermal insulator, are less<br />

important and do not aect signicantly the global dynamic behavior of the support.<br />

So, they will be ignored in this model.<br />

At this level, the precise acceleration at which the support is submitted, is not known<br />

(because a model of the entire satellite is not available). In addition, the boundary conditions<br />

can not be determined exactly. So, it was decided to follow a conservative approach<br />

using the following approximations:<br />

• The connections with the endless screws are supposed to be perfectly rigid in the 3<br />

spatial directions. This approximation is pretty good because the spacers between<br />

the electronic cards and the support can be considered as extremely rigid in their<br />

longitudinal axis. In addition, the endless screws, which are xed at each of their<br />

extremities, can also be considered as rigid xations. However, for the rotational<br />

Degrees of Freedom (DoF), there is a certain exibility, due to the little set between<br />

the screws and the support's holes, that can not be measured accurately without<br />

having a complete engineering model of the satellite. So, these ones are let free,<br />

which lead to an overestimation of the structural deections and an underestimation<br />

of the natural frequencies.<br />

• The acceleration applied to the support is not known. However, a known value is the<br />

maximal acceleration of the launcher along its longitudinal axis, which is the more<br />

constraining one and rates 6.3 g. So, using a SF of 2, this acceleration reaches a<br />

value of 12.6 g that will be applied to our model.<br />

The results obtained are presented in Figure 3.24. It can be directly noted that the<br />

maximum Von Mises stress inside the support is situated, as predicted before, at the level<br />

of the legs. However, even with our approximations and the conservative approach applied<br />

here, the value of this stress (1.53 MP a) is strongly lower than the yield stress of the<br />

Al − 7075 T 6, which has a value of 444.5 MP a.<br />

66


Figure 3.24: Maximal Von Mises stress inside the support<br />

67


3.9.3 Dynamic behavior<br />

Finally, it is also important to verify that the dynamic behavior of the support could<br />

not cause trouble for the satellite.<br />

The modeling of the support is the same as the one described in the previous section.<br />

The 5 rst natural frequencies obtained are available in Table 3.7.<br />

Frequency (Hz)<br />

1 2869.53<br />

2 3219.66<br />

3 5978.78<br />

4 6022.2<br />

5 6192.42<br />

Table 3.7: First natural frequencies of the battery support<br />

The two most signicant modes are represented in Figure 3.25 to illustrate the general<br />

dynamic behavior of the assembly.<br />

From these results, it can be deduced that the dynamic behavior of the support will<br />

not aect the global behavior of the satellite in a critical way. It can also be noted that the<br />

second vibration mode of the support is a bending mode of the legs. This fact consolidates<br />

our idea that the principal stress concentration will be situated at this particular place,<br />

and validates our choice to increase the thickness of these legs.<br />

To conclude, these results demonstrate that our design is valid with regard to the<br />

dynamic behavior.<br />

68


Figure 3.25: The two most signicant modes of the battery support<br />

69


3.10 Summary<br />

Through this chapter, the design of a new support for the batteries was discussed.<br />

First, the problems encountered with the last year's design were exposed to explain<br />

why this particular design had to be given up.<br />

Then, the available mass budget to realize this support was calculated, which is a crucial<br />

step when a new design had to be created.<br />

The selection of the several components which have to be included inside the support,<br />

was also exposed. In addition, the material selection, which has to be in accordance with<br />

several requirements that were listed, was studied.<br />

Finally, the design procedure was presented step by step, just like it progressed during<br />

the whole year. This design was then validated by several analysis and calculations.<br />

At the moment where this thesis ends, the plans are provided to the support's manufacturers<br />

(a rst rough shape of the box is presented in Figure 3.26).<br />

Figure 3.26: A rst rough shape of the battery support<br />

70


Thermal tests under vacuum conditions and structural ones of this support are planned<br />

for June 2010. For the STRU subsystem, the following verications will have to be performed:<br />

• A visual inspection of the support to detect potential failures.<br />

• A sinusoidal test before and after the thermal tests. This particular test allows to<br />

compare the natural frequencies of the support before and after it is submitted to<br />

vacuum conditions. If there is any dierence between these two sets of frequencies, it<br />

means that the support suers some failures (which could not be detected visually)<br />

and that its structural integrity is not guaranteed in the harsh space environment.<br />

It should be noted that, between these two sinusoidal tests, the support must not be<br />

modied in any way.<br />

• A verication of the tting torques applied to the screws.<br />

71


Chapter 4<br />

Electronic cards dynamic modeling<br />

4.1 Introduction<br />

One of the most vulnerable part of a satellite is its electronic cards. During launch,<br />

the electronic components are subjected to harsh random vibration environment and some<br />

failures can occured. The probability of these failures is strongly inuenced by the local<br />

PCB response. To prevent from this undesirable phenomenon and to avoid it, it is a current<br />

practice to create FE models of electronic cards to predict their local vibration response.<br />

Throughout this chapter, the problem of how to model the dynamic behavior of an<br />

electronic card, will be discussed.<br />

Firstly, we will go around the state-of-the-art in this domain. Multiple methods will be<br />

presented and their accuracy will be briey discussed.<br />

Then, these methods will be applied to a practical example: the homemade OBC (OBC<br />

2) of <strong>OUFTI</strong>-1.<br />

Finally, an experimental test will be performed and the results will be correlated with<br />

the ones obtained by the FE model. The inuence of various parameters on the PCB<br />

response will also be studied through some sensitivity analysis.<br />

4.2 State-of-the-art<br />

It should be noted that the information incorporated in this section come from references<br />

[44] to [49] (especially from references [44] and [45]).<br />

72


4.2.1 Accuracy of the model<br />

First and foremost, it is important to note that the accuracy of the FE model which<br />

will be created, is strongly related to the decision-making. Indeed, if the results required<br />

are the component internal stresses, a detailed FE model is necessary. However, the long<br />

time required to build and solve this type of model can not be justied when only the<br />

PCB response is required. In this case, it is possible to greatly simplify the model by using<br />

several methods which will be discussed later.<br />

In addition, the accuracy of a FE model will be mainly dependent on various sources<br />

of error, including:<br />

• Manufacturing variability, which will cause deviations in the vibration response of<br />

supposedly identical PCBs. This variability includes not only material and assembly<br />

properties, but also dimensional tolerances applied during the manufacturing<br />

procedure.<br />

• Inaccuracies in the denition of the model input parameters. They can result from<br />

the modeling assumptions used or the impossibility to obtain reliable values of these<br />

parameters (no prototype available, ...).<br />

• Errors in the solution process (e.g., linear solutions in non-linear situations).<br />

The major diculties encountered during the creation of a PCB FE model are so:<br />

• To specify the input parameters, namely: stiness, density, boundary conditions and<br />

damping. The accuracy with which these parameters are specied, will determine<br />

the accuracy of the predicted response. So, experimental tests must be performed<br />

to determine reliable values of these parameters in order to obtain an accurate FE<br />

model.<br />

• Variation of these parameters due to manufacturing, assembly, wear, ...<br />

An example of manufacturing variation is shown in Figure 4.1, where the same test is<br />

performed on supposedly identical PCBs. It can be directly seen than the responses<br />

obtained are really dierent one from the other.<br />

73


Figure 4.1: Example of manufacturing variation [44]<br />

• Limitations due to parameters and variation sources which can not be easily specied,<br />

such as non-linear eects, limitations of the FE mesh, physical eects that are too<br />

complicated to be easily included in the model (i.e., air or acoustic inuences), ...<br />

Most often, these diculties are overcome by including appropriate SF in the model.<br />

4.2.2 Creating FE models of PCBs<br />

The procedure to follow for creating FE models of PCBs, can be divided into 5 parts:<br />

• Determination of the PCB properties<br />

• Recognition of components eects<br />

• Modeling of the chassis<br />

• Denition of the boundary conditions<br />

• Introduction of damping<br />

4.2.3 Determination of the PCB properties<br />

The PCB properties include: Young's modulus, Poisson ratio, density, thickness, ...<br />

The specication of these properties represents the most dicult step in creating a PCB<br />

FE model. In general, their values may be provided by manufacturers. However, these<br />

values are not always exact (as shown in Figure 4.2). So, it is better to determine them<br />

by ourselves.<br />

74


Figure 4.2: Example of dierences between the values provided by manufacturers and the<br />

values determined by experimental tests [44]<br />

The Young's modulus can be determined using a static bend test. However, it is important<br />

to note that most PCBs are laminated and so, their properties may vary depending<br />

on the direction of loading. Thus, the determination of the Young's modulus necessitates<br />

three dierent tests, one according to each axis of the PCB. Figure 4.3 shows an example<br />

of this type of test.<br />

Figure 4.3: Illustration of the experimental set-up for a static bend test<br />

The shear modulus of a PCB may most conveniently be determined through a torsion<br />

test, as shown in Figure 4.4.<br />

So, the shear modulus can be calculated using Equation 4.2.1:<br />

G xy = 3 K t a b<br />

4 t 3 (4.2.1)<br />

where K t is the slope of the load displacement curve, a and b are the dimensions of the<br />

specimen edges and t is its thickness.<br />

75


Figure 4.4: Illustration of the experimental set-up for a torsion test [44]<br />

If the PCB mass and dimensions are known, the density of the PCB is easily found<br />

using Equation 4.2.2.<br />

ρ P CB = m P CB<br />

V P CB<br />

(4.2.2)<br />

where m P CB is the mass of the PCB and V P CB is its volume. It is important to note that<br />

this equation is based on the implicit assumption that the thickness is constant over the<br />

entire PCB (which it is not always the case as it can be observed in Figure 4.2).<br />

An other important fact to keep in mind is that all the tests must be performed on the<br />

etched PCB because, as it was already mentioned, manufacturing will induce changes in<br />

the PCB properties, like loss of mass or stiness.<br />

4.2.4 Choice of mesh element<br />

After the denition of the PCB properties, it remains to choose the type of mesh<br />

elements which will be used to mesh the PCB. In accordance with the particular geometry<br />

of this type of structure, two possibilities can be pointed up:<br />

• Volume elements<br />

• Shell elements<br />

In this particular case, the shell elements are the best solution because a PCB can be<br />

linked to a thin structure. Indeed, these elements allow to obtain results as accurate as if<br />

volume elements were used, but they require less computation time and memory, which is<br />

a great advantage in the current world of industry.<br />

To convince ourselves of this fact, let us use the two solutions on a simple example.<br />

Consider a steel plate with an area of 10×10 cm. The properties of this particular material<br />

are well known:<br />

76


• Young's modulus: E = 205 GP a<br />

• Poisson ratio: ν = 0.3<br />

• Density: ρ = 7850 kg/m 3<br />

Consider also that the plate has a constant thickness of 1.6 mm (which is the given<br />

thickness of <strong>OUFTI</strong>-1 PCBs). This plate was meshed once with volume elements and<br />

once with shell elements. Then, the two sets of results are correlated. The most popular<br />

correlation techniques make use of natural frequencies ω and mode shapes φ. And, the<br />

most popular indicators of the correlation between two sets of data are:<br />

• The frequency deviation, that can be calculated using Equation 4.2.3.<br />

This indicator is generally expressed in %.<br />

∆f (ω 1 , ω 2 ) = |ω 1 − ω 2 |<br />

ω 1<br />

(4.2.3)<br />

• The Modal Assurance Criterion (MAC), that can be calculated using Equation 4.2.4.<br />

MAC (φ 1 , φ 2 ) =<br />

(<br />

φ<br />

T<br />

1 φ 2<br />

) 2<br />

(φ T 1 φ 1 ) (φ T 2 φ 2 )<br />

(4.2.4)<br />

MAC values oscillate between 0 and 1, a unitary value meaning perfect correlation.<br />

The results obtained by using these two indicators in our particular case, are presented<br />

in Table 4.1 and in Figure 4.5.<br />

Natural frequencies obtained Natural frequencies obtained Frequency<br />

using volume elements (Hz) using shell elements (Hz) deviations (%)<br />

1 528.2 527.8 0.08<br />

2 771 770.2 0.1<br />

3 954.8 953.8 0.1<br />

4 1364.1 1361.9 0.16<br />

5 1364.1 1361.9 0.16<br />

Table 4.1: Frequency deviations between the two models<br />

77


Figure 4.5: MAC between the two models<br />

According to these results, it can be armed that the shell elements allow to obtain<br />

results as accurate as the ones obtained using volume elements. The slight dierence in<br />

the MAC matrix is only due to the fact that the fourth and fth modes of the plate are<br />

symmetrical ones, as it is shown in Figure 4.6.<br />

Figure 4.6: Symmetrical modes of the steel plate<br />

So, from this point, the shell elements will be used to mesh the electronic cards during<br />

FE analysis until the end of this thesis.<br />

78


4.2.5 Recognition of components eects<br />

When components are soldered to a PCB, they locally increase the mass and stiness of<br />

this PCB. So, to be accurate, especially when a large number of components are present,<br />

the FE model of the PCB must include these eects.<br />

Theoretically, this problem can be solved by creating a detailed FE model of each<br />

individual component present on the card, as illustrated in Figure 4.7. However, as already<br />

mentioned in section 4.2.1, this solution, which brings a high level of detail in the model,<br />

is not justied if only the PCB response is required. Indeed, in this type of model, each<br />

lead of the component must be modeled using three rotational stiness elements, for which<br />

the stiness must be determined by a trial/error approach. This requires too much time<br />

and eorts for the expected results and so, this type of modeling is generally less favorable<br />

than simplied methods.<br />

Figure 4.7: Detailed model of a component [45]<br />

Dierent levels of simplication are possible. They are summarized here:<br />

• Simple method: The components eects are completely neglected. The reasoning<br />

behind this method is the following one: ignoring the stiness increases the PCB<br />

response and decreases the natural frequencies, whilst ignoring the mass decreases the<br />

PCB response and increases the natural frequencies. So, each eect "compensates"<br />

the other. This method is useful when no data about the components properties and<br />

location are available.<br />

• Global mass smearing method: The mass of each component is spread out over<br />

the entire area of the PCB. The stiness contributions are completely neglected. This<br />

method, which could be appeared not accurate, is based on the fact that neglecting<br />

the stiness is a conservative approach (for the reasons exposed in the previous point).<br />

To apply this method to a particular model, the card must be weighted and then,<br />

its mass must be divided by the volume of its PCB to obtain the global density to<br />

apply on this one.<br />

79


• Global mass/stiness smearing method: Both mass and stiness of each component<br />

are spread out over the entire area of the PCB. To apply this method to a<br />

particular model, the same calculation than in the previous point, must be performed<br />

to obtain the global density. The global stiness is achieved using an area weighted<br />

average of the locally smeared Young's moduli as shown in Equation 4.2.5.<br />

E global =<br />

∑<br />

Ei A i<br />

A total<br />

(4.2.5)<br />

where E i is the Young's modulus at the location i on the PCB, A i is the surface area<br />

of this location and A total is the total surface area of the PCB.<br />

• Local smearing method: This method is the same as the previous one, except<br />

that, instead of smearing the components properties over the entire PCB, they are<br />

spread out over local regions of the PCB, where the local regions can be dened as<br />

the projection of the components area on the PCB.<br />

These methods are illustrated in Figure 4.8.<br />

Figure 4.8: Illustration of the dierent simplication methods [45]<br />

80


Now, the question is: When each method can be applied and why <br />

To answer to this question, it should be noted that electronic components can be divided<br />

into three categories:<br />

• Light components: This category includes small discrete components, such as<br />

resistors or transistors. The mass and stiness of these particular components are so<br />

small than they can be neglected. So, in general, the simple method is applied to<br />

these components.<br />

• Surface Mount Technology (SMT) components: This category symbolizes<br />

components such as Quad Flat Pack (QFP) Ball Grid Array (BGA) and Pin Grid<br />

Array (PGA), which are generally about 10 − 30 mm square. The mass and stiness<br />

increases of these particular components are proportional to their length and inversely<br />

proportional to the thickness of the PCB on which they are soldered. So, in general,<br />

the global or local smearing methods are applied to these components.<br />

• Heavy components: This category includes large components, such as transformers,<br />

large power capacitors and resistors. Due to their important mass and stiness,<br />

these components are generally modeled using a local smearing method or, more<br />

rarely, a detailed FE model.<br />

These categories are illustrated in Figure 4.9.<br />

Figure 4.9: Illustration of the dierent categories of components [45]<br />

81


4.2.6 Modeling of the chassis<br />

A general rule of thumb when a FE model is created, is to always model the next level<br />

up from the structure. In the case of PCBs modeling, this level is the chassis.<br />

If the chassis is not included in the model, it may severely aect the accuracy of this one,<br />

unless the chassis will be extremely rigid in comparison to the PCB. Indeed, in reality, the<br />

PCB is xed on the chassis. So, to obtain accurate results, it is a current practice to attach<br />

the PCB on this structure during experimental tests. Thus, the boundary conditions are<br />

identical to the ones encountered during the lifetime of the PCB and the obtained dynamic<br />

response is more accurate.<br />

4.2.7 Denition of the boundary conditions<br />

Once the FE models of the PCB and the chassis are created, the next step is to<br />

combine them together. To achieve this, rigid and/or exible FE assembly methods are<br />

used, depending on the studied case:<br />

• In terms of translational displacement, most xing methods are very sti. So, they<br />

can be modeled using rigid assembly methods.<br />

• In terms of rotational displacement, all xing methods display some exibility. So,<br />

the use of rotational local stiness assembly methods, is required.<br />

In this last case, the main diculty is to determine the value of the rotational stiness.<br />

Two situations are possible:<br />

• No prototype of the PCB is available. In this case, the options are very limited: the<br />

value of this stiness can be estimated basing on subjective experience, or a detailed<br />

FE model of the joint must be created (which takes, as already mentioned, too much<br />

time in several cases).<br />

• A prototype of the PCB is available. In this case, tests can be performed and the<br />

value of the stiness can be determined using a trial/error approach. However, it<br />

is important to note that methods developed in some works [50], were originally<br />

intended for use with card-lock style xing mechanisms, which provide clamping<br />

force along the entire edge of a PCB. So, these methods should be used with caution<br />

in the case of locally xed PCBs (e.g., bolted, ...).<br />

4.2.8 Introduction of damping<br />

Several methods exist to measure experimentally the damping of a given structure (see<br />

references [51] to [53]):<br />

82


• The logarithmic decrement method: This method is a time domain approach.<br />

It calculates damping based on the free decay of oscillations in the structure using<br />

Equation 4.2.6.<br />

( )<br />

xn<br />

∆ = ln = 2 π ζ<br />

x n+1<br />

√ ⇒ ζ = 1<br />

√<br />

1 − ζ<br />

2<br />

1 + ( 2 π<br />

∆<br />

) 2<br />

(4.2.6)<br />

where ∆ is the logarithmic decrement between times n and n + 1, x i is the response<br />

of the structure at a given time i and ζ is the damping factor.<br />

If the assumption of low damping can be applied, this equation is greatly simplied.<br />

This simplication is presented in Equation 4.2.7.<br />

∆ ≈ 2 π ζ ⇒ ζ = 2 π<br />

∆<br />

(4.2.7)<br />

• The peak-amplitude method: This method is a frequency domain approach. It<br />

calculates damping based on the Frequency Response Function (FRF) of the structure.<br />

Once the FRF is obtained, the response at the half-power points (illustrated in<br />

Figure 4.10) is measured and the damping can be calculated as shown in Equation<br />

4.2.8.<br />

Figure 4.10: Illustration of the peak-amplitude method [53]<br />

83


Q =<br />

ω k<br />

= 1<br />

ω b − ω a 2 ζ ⇒ ζ = 1 ω b − ω a<br />

(4.2.8)<br />

2 ω k<br />

where Q is the quality factor, ω r is the natural frequency, ω a and ω b are the frequencies<br />

of half-power points and ζ is the damping factor.<br />

• The Steinberg's equation: This method is an analytical one [54]. It states that<br />

the transmissibility at resonance of a PCB is equal to several times (depending of<br />

the natural frequency considered) the square root of the natural frequency, as shown<br />

in Equation 4.2.9.<br />

Q = a √ ω r<br />

with<br />

⎧<br />

⎪⎨<br />

⎪⎩<br />

a = 0.5<br />

ω r ≤ 100 Hz<br />

a = 0.75 100 Hz < ω r ≤ 200 Hz<br />

(4.2.9)<br />

a = 1 200 Hz < ω r ≤ 400 Hz<br />

a = 2 400 Hz < ω r<br />

where Q is the transmissibility at resonance and a is the tting factor based on ω r ,<br />

which is the natural frequency. Unfortunately, the data on which the Steinberg's<br />

method is based, are unavailable and therefore, the method is unveriable. For this<br />

reason, it will not be used in this thesis.<br />

• And a lot of other methods: Circle tting, Stochastic Subspace Identication (SSI),<br />

Ibrahim Time Domain (ITD), ...<br />

It is important to note that, in case of low damping, the frequency domain methods are<br />

pretty poor because the curves on which they are based, are dicult to measure accurately<br />

due to the high rate of change of the frequency response curve. So, in case of low damping,<br />

time domain methods will be preferred.<br />

An other important remark is that damping may vary with the excitation level. So,<br />

several tests must be performed at several levels of excitation to determine an accurate<br />

value of the damping factor.<br />

4.2.9 Dynamic response computation<br />

To avoid large errors when computing the dynamic response of a PCB, two points<br />

should be considered:<br />

• Enough modes should be computed in the modal solution to excite a signicant<br />

fraction (at least 90%) of the total mass of the structure.<br />

• If the deection of the PCB is comparable to its thickness, a non-linear analysis will<br />

be preferred.<br />

84


4.3 Application to the homemade on-board computer of<br />

<strong>OUFTI</strong>-1<br />

Now, the several methods discussed through section 4.2 will be applied to the particular<br />

case of the homemade OBC of <strong>OUFTI</strong>-1. This electronic card is located just above the<br />

FM 430 card from Pumpkin (as shown in Figure 4.11).<br />

Figure 4.11: Location of the OBC 2 card in <strong>OUFTI</strong>-1<br />

4.3.1 Denition of the PCB properties<br />

Each electronic card of <strong>OUFTI</strong>-1 was manufactured by Deltatec [55], a very high-level<br />

design house specialized in advanced technologies. They provided us the properties of the<br />

FR 4 material they used to manufacture the PCB of the <strong>OUFTI</strong>-1 electronic cards. These<br />

properties are presented in Table 4.2.<br />

Properties Values<br />

Young's modulus 20 GP a<br />

Poisson ratio 0.3<br />

Density 2200 kg/m 3<br />

Ultimate stress 423 MP a<br />

Table 4.2: Properties of the FR 4 material<br />

85


Their study was based on the fact that the material is isotropic. They did not determine<br />

the anisotropic values because, owing to the PCB assembly (which is a piling up of several<br />

layers of FR 4 and copper), these values are too complicated to obtain. So, anisotropic FE<br />

model of the PCB can not be performed in this thesis.<br />

Finally, they give us a value of 1.6 mm for the PCB thickness. But, measuring it by<br />

ourselves, the values obtained are included in a range of: [1.65 − 1.75 mm]. So, the value<br />

which will be imposed to our FE models, is 1.7 mm.<br />

The mesh elements that will be used to mesh the PCB of the OBC 2 card, as already<br />

mentioned in section 4.2.4, are shell elements.<br />

Finally, for the damping, according to reference [49], a modal damping of 2% is a quite<br />

representative value for a PCB because there is a great friction between the dierent layers<br />

which constitute it. In our case, a value of 1% will be applied to the model to keep a<br />

security margin of 1%.<br />

4.3.2 Recognition of the components eects<br />

First of all, we have to decide which type of components will be taken into account<br />

in the FE model of the OBC 2. As already mentioned, these components can be divided<br />

into three dierent categories: light, SMT and heavy components. The classication of<br />

components soldered to this card, is presented in Figure 4.12.<br />

Figure 4.12: Components classication of the OBC 2 card<br />

86


In our case, only the SMT and heavy components will be included in the FE model of<br />

the OBC 2 card. The light components, due to their negligible eect on the stiness and<br />

mass properties of the PCB, will not be modeled in any application. The Catia modeling<br />

of the OBC 2 card is presented in Figure 4.13.<br />

Figure 4.13: Catia modeling of the OBC 2 card<br />

To take into account the mass contributions of these components on the OBC 2 response,<br />

the methods exposed in section 4.2.5 will be applied one by one. Thus, the accuracy<br />

of each of them will be discussed, based on experimental results obtained through the<br />

test that will be presented in section 4.3.3.<br />

The stiness contributions will not be included in FE models for several reasons:<br />

• First, the datasheets provided with most electronic components, do not contain this<br />

information.<br />

• Then, realizing a static bend test on each component is too expensive with regard to<br />

the expected results.<br />

• Finally, as already mentioned, ignoring the stiness is a conservative approach. So,<br />

the natural frequencies that will be obtained, will be lower than the real ones. This<br />

fact is not critical in our case because, in general, LV manuals specify minimum<br />

values for the payload natural (fundamental) frequency of vibration in order to avoid<br />

dynamic coupling between low frequency dynamics of the LV and payload modes.<br />

So, if the natural frequencies obtained using this method fulll the requirements, the<br />

real ones will automatically fulll these requirements. An example of this type of<br />

requirement is presented in Figure 4.14.<br />

87


Figure 4.14: Example of spacecraft stiness requirements for dierent launchers<br />

4.3.3 Experimental test of the OBC 2 card<br />

In this section, the description of the experimental test performed on the OBC 2 card,<br />

will be explained. Three main parameters have to be dened for a vibration test: the<br />

boundary conditions (the support), the type of excitation and the acquisition system.<br />

Each of these parameters has a great inuence on the results obtained and so, they have<br />

to be chosen carefully.<br />

Boundary conditions<br />

The choice of the support used during a test is strongly related to the type of excitation<br />

and to the acquisition system. Indeed, some system (e.g., laser transducers) needs an<br />

accurate pointing and so, a rigid support to maintain the tested structure. So, this choice<br />

has to be made in compliance with the other requirements of the test.<br />

In our case, the test was realized in "free-free" boundary conditions. This particular<br />

conguration was chosen in order to avoid the apparition of additional unknowns concerning<br />

the xing method. In addition, it gives rigid body modes at low frequencies, which<br />

allow to distinguish them from the real modes of the OBC 2 card. To simulate these conditions,<br />

the card was suspended by four elastics attached to the four endless screws' holes<br />

of the card.<br />

Type of excitation<br />

Then, the type of excitation that will be used, must be dened. The selection criteria<br />

are generally: simplicity and reliability.<br />

In our case, the hammer excitation was chosen. Indeed, owing to the restricted dimensions<br />

of the electronic card, this type of excitation is the most practical one. The<br />

hammer used is the smallest available in the laboratory. It was equipped with a tip made<br />

of vinyl, which allow to obtain an acceptable Power Spectral Density (PSD) in the range<br />

of: [0 − 2000 Hz]. A better choice, with regard to the PSD, is to use the stainless steel<br />

88


tip, but this one is subject to the "bounce" eect, which can cause perturbations on the<br />

vibration response of the structure. No additional mass was used to increase the hammer<br />

impact. Note that an exponential window was applied to the excitation signal. This is a<br />

current practice when a hammer is used to perform the test.<br />

It should also be noted that no impact averaging was performed. Indeed, to apply this<br />

method, impacts include in a series must be the same and, any dierence in the impact<br />

location or intensity has a signicant eect on the results. Now, the OBC 2 card is so<br />

small that the reproduction of the same impact several time can not be assured. In our<br />

case, the excitation point was chosen on a corner of the card (see Figure 4.15).<br />

Acquisition system<br />

Finally, the acquisition system, which takes measurements on the structure, has to be<br />

chosen. An important requirement, that will be discussed in section 4.3.8, is that this<br />

acquisition system does not interfere with the dynamic response of the structure.<br />

In our case, it was decided to use an accelerometer. This type of sensor is currently<br />

used when the signal to measure is a high-frequency one (e.g., a shock), which is the case<br />

here. Once again, owing to the restricted dimensions and mass of the electronic card, this<br />

accelerometer has to be as lightweight and small as possible, to avoid any interference with<br />

the OBC 2 dynamic response. In our case, the accelerometer used weights only 0.2 g and<br />

is enough small for being placed everywhere on the electronic card.<br />

It was chosen to realize 22 measurements on the card. According to this particular<br />

choice, the best solution to attach the accelerometer is the beeswax, which allows a good<br />

grip without interfering with the dynamic response, and an easy handling. The measurement<br />

points are presented in Figure 4.15.<br />

89


Figure 4.15: Location of measurement and excitation points<br />

The acquisition parameters applied to the measured signal, are:<br />

• A frequency sampling of 512 Hz<br />

• A resolution of 2.5 Hz<br />

The complete set-up is presented in Figure 4.16.<br />

Figure 4.16: Experimental set-up for the test of the OBC 2 card<br />

90


The global FRF of the OBC 2 card is presented in Figure 4.17.<br />

Figure 4.17: Global FRF of the OBC 2 card<br />

Now that everything was dened, the dierent simplication methods can be applied<br />

to the OBC 2 card and compared to experimental results. This comparison will be realized<br />

using the two indicators introduced in section 4.2.4, namely the frequency deviation and<br />

the MAC.<br />

4.3.4 Application of the simple method<br />

The rst method to be applied is the simple method. So, all the components eects are<br />

completely neglected and only the PCB remains. These properties being already dened,<br />

the results can be directly computed. They are presented in Table 4.3 and in Figure 4.18.<br />

Corresponding modes Frequency deviations (%)<br />

1 1.01<br />

2 11.37<br />

Table 4.3: Frequency deviations between corresponding modes using the simple method<br />

91


Figure 4.18: MAC matrix using the simple method<br />

It can be immediately noted that these results are not good because only two modes<br />

of the card are found by the model.<br />

This great dierence between the results from the model and the ones obtained by<br />

experimental test, can be explained by several facts:<br />

• Firstly, the boundary conditions are not fully "free-free" in the experimental set-up<br />

(this type of boundary conditions is impossible to reproduce in reality owing to the<br />

gravity). This could bring some unknowns about these boundary conditions which<br />

are not taken into account in the FE model. This fact could lead to dierent dynamic<br />

responses of the PCB.<br />

• Then, the choice of hammer and/or tip could be not appropriate to our case. And, if<br />

the structure is not well excited, its dynamic response can not be measured accurately.<br />

• Finally, owing to the restricted dimensions of the studied structure, the exact location<br />

of measurement points is uncertain. Any error in this location could lead to a<br />

comparison between two points (the measurement point and its "equivalent" from<br />

the FE model) that have fundamentally dierent displacements.<br />

But, for our study, the results obtained are sucient and so, an other test will not be<br />

performed.<br />

92


4.3.5 Application of the global mass smearing method<br />

The second method to be applied is the global mass smearing method. In this method,<br />

the mass of each component is spread out over the entire area of the PCB. So, the global<br />

density is obtained by dividing the mass of the complete electronic card by the volume of<br />

the PCB. This calculation is realized in Equation 4.3.1.<br />

ρ global = m OBC 2 50.4607 × 10−3<br />

= = 3536.1388 kg/m 3 (4.3.1)<br />

V P CB 1.427 × 10 −5<br />

The results are presented in Table 4.4 and in Figure 4.19.<br />

Corresponding modes Frequency deviations (%)<br />

1 21.89<br />

2 30.07<br />

Table 4.4: Frequency deviations between corresponding modes using the global mass smearing<br />

method<br />

Figure 4.19: MAC matrix using the global mass smearing method<br />

It can be remarked that the results are better than the ones obtained using the simple<br />

method. So, this method is more accurate than the preceding one (which is in accordance<br />

with the theory).<br />

93


4.3.6 Application of the local mass smearing method<br />

The third method to be applied is the local mass smearing method. In this method,<br />

the mass of each component is spread out over local regions of the PCB. So, the density<br />

of the PCB is locally increased. The value of this increase can be calculated by dividing<br />

the mass of each component by the product between the base area of the component and<br />

the thickness of the PCB. The increase of density caused by each component is presented<br />

in Table 4.5.<br />

Components Equivalent densities (kg/m 3 )<br />

P C 104 connector 15308.5<br />

I 2 C − EEP ROM 2717.2<br />

MAX 890 2335.3<br />

MSP 430 2068.2<br />

Crystal 32.768 kHz 3816.4<br />

JT AG connector 4275.9<br />

Table 4.5: Increases of density due to each component of the OBC 2 card<br />

The results are presented in Table 4.6 and in Figure 4.20.<br />

Corresponding modes Frequency deviations (%)<br />

1 21.46<br />

2 25.98<br />

Table 4.6: Frequency deviations between corresponding modes using the local mass smearing<br />

method<br />

Once again, it can be remarked that the results are better than the ones obtained using<br />

the global mass smearing method. So, this method is more accurate than the preceding<br />

one (which is in accordance with the theory).<br />

94


Figure 4.20: MAC matrix using the local mass smearing method<br />

4.3.7 Application of a homemade method<br />

Generally, the methods used in the previous sections, are applied to the entire PCB.<br />

However, as already mentioned in section 4.2.5, these methods can be particularized to the<br />

dierent categories of components and several methods can be applied on the same PCB.<br />

The homemade method described here is based on this concept. It considers the following<br />

simplications:<br />

• The light components are completely neglected.<br />

• The SMT components, including: the I 2 C − EEP ROM, the two MAX 890, the<br />

MSP 430, the Crystal 32.768 kHz and the JT AG connector, are modeled using the<br />

global mass smearing method. This solution was chosen because it is easier to apply<br />

than the local mass smearing method and, with regard to the mass contributions of<br />

these components, it brings the same level of accuracy.<br />

• The heavy components, including only the P C 104 connector, are modeled using a<br />

detailed FE model. This solution was chosen because, owing to its dimensions and<br />

its properties, the P C 104 connector plays an important role in the PCB dynamic<br />

response. So, it is better to model it entirely. To be able to create this particular<br />

model, the properties of this connector are necessary. Its mass is easily obtained<br />

by weighting it and has a value of 10.12 g, which leads to a density calculated in<br />

Equation 4.3.2<br />

95


ρ P C 104 connector =<br />

10.12 × 10−3<br />

7.193 × 10 −6 = 1406.9234 kg/m3 (4.3.2)<br />

For its stiness, no static bend test was possible to perform. So, its value was xed<br />

based on reference [49], where the Young's modulus of an equivalent component was<br />

determined by test. So, the Young's modulus used in our FE model is 846 MP a.<br />

Applying this simplications on the FE model, the results obtained are presented in<br />

Table 4.7 and in Figure 4.21.<br />

Corresponding modes Frequency deviations (%)<br />

1 15.29<br />

2 17.33<br />

Table 4.7:<br />

method<br />

Frequency deviations between corresponding modes using the homemade<br />

Figure 4.21: MAC matrix using the homemade method<br />

It can be remarked that this particular method is the one which gives the best results<br />

according to the MAC and the frequency deviations. For these last ones, the great<br />

dierences between the values obtained by the model and the ones resulting from the experimental<br />

test, can be easily explained. They are due to the fact that the stiness is<br />

ignored, and that ignoring the stiness is a conservative approach which underestimates<br />

the values of the natural frequencies. So, in the following chapter, this particular method<br />

will be applied to each electronic card of <strong>OUFTI</strong>-1 to obtain its complete FE model.<br />

96


4.3.8 Sensitivity analysis<br />

During the description of the modeling strategy, it was highlighted that several parameters<br />

play an important role on the accuracy of the obtained FE model. In this section, the<br />

eect that a variation of some of these parameters causes on the PCB dynamic response,<br />

will be studied.<br />

Mass of the sensor used during the experimental test<br />

The rst analysis concerns the mass of the sensor used to realize the measurements<br />

during the test. Indeed, this sensor brings a local increase of mass and stiness on the<br />

PCB, just as each electronic component. To obtain accurate results, this fact has normally<br />

to be taken into account in our model, which is never the case when a FE model is realized.<br />

So, the question is: what is the real eect of the sensor on the PCB dynamic response <br />

To try to answer to this question, a sensitivity analysis will be performed by integrating<br />

sensors of several mass in our model. Then, the shape of the fourth mode of the OBC 2<br />

card will be studied for 3 cases: a sensor of 0.2 g (as the sensor used in our case), one of<br />

1 g and the last of 2 g.<br />

To realize this, a concentrate mass is added to the model. However, this mass has to<br />

be placed at a strategic point to obtain results which highlight its inuence on the mode<br />

shape. Indeed, if this mass was placed on a vibration node, nothing could be observed<br />

because it would not be excited. So, to realize this choice, the initial shape of the fourth<br />

mode is necessary. This one is given in Figure 4.22.<br />

Figure 4.22: Initial shape of the OBC 2 card's fourth mode<br />

97


According to this particular shape, it was chosen to place the sensor at the level of the<br />

measurement point 1 (see Figure 4.15), which is situated in one of the two corners on the<br />

opposite side of the P C 104 connector, because it is this part of the electronic card which<br />

presents the greatest deformations and so, which participates the most to this mode. Table<br />

4.8 presents the natural frequencies obtained in each case and their deviations with regard<br />

to the initial one, and Figure 4.23 shows the eect on the mode shape.<br />

Cases Natural frequencies (Hz) Frequency deviations (%)<br />

Initial 907.68 /<br />

Sensor of 0.2 g 893.81 1.53<br />

Sensor of 1 g 862.66 4.96<br />

Sensor of 2 g 847.62 6.62<br />

Table 4.8: Natural frequencies obtained by adding sensors of several mass to the FE model<br />

Figure 4.23: Mode shapes obtained by adding sensors of several mass to the FE model<br />

98


Thanks to these results, the following conclusions can be drawn:<br />

• The increase of mass brought by the sensor has an important eect on the value of<br />

the natural frequencies. Indeed, it can be seen that, even with a sensor of 0.2 g,<br />

the shift in frequency reaches a value of 13.87 Hz, which is not negligible. And this<br />

eect becomes more important when the mass of the sensor increases. This problem<br />

is essentially due to the fact that the studied structure has very restricted dimensions<br />

and mass. So, any increase of mass, so small it is, involves interferences with the<br />

PCB dynamic response.<br />

• This eect can also be highlighted with the mode shape. Indeed, it can be seen that<br />

the amplitude of the PCB deformation at the sensor location decreases when the<br />

mass of the sensor increases.<br />

So, in this particular case, it will be better to use a sensor which does not need to<br />

touch the structure to perform the measurements (e.g., laser transducers, ...). This allows<br />

to avoid any interference with the dynamic response of this structure.<br />

Modeling of the P C 104 connector<br />

The second analysis concerns the modeling of the P C 104 connector. It was choose to<br />

represent it using a detailed FE model, for reasons exposed previously. But, the question<br />

is: what does this particular choice involve on the PCB dynamic response To answer to<br />

this question, the fourth mode of the OBC 2 card for the two dierent type of modeling,<br />

is shown in Figure 4.24.<br />

Figure 4.24: Inuence of a change of modeling strategy for the P C 104 connector on the<br />

OBC 2 card's fourth mode<br />

Thanks to these results, it can be seen that the inuence of the P C 104 connector on<br />

the PCB dynamic response is stronger using the detailed FE model (the deformation of the<br />

card is less important at the connector location and more important on the opposite side).<br />

99


This is due to the fact that this connector brings a non-negligible increase of stiness to<br />

the PCB, and that this particular increase is not taken into account using the local mass<br />

smearing method. So, the best solution is well to use the detailed FE model for the P C 104<br />

connector.<br />

4.4 Summary<br />

Through this chapter, the electronic cards dynamic modeling was discussed.<br />

First, we took a turn of the several existing methods to realize this particular modeling.<br />

The accuracy of each method was studied and the critical parameters on which this<br />

accuracy is mainly dependent, were highlighted.<br />

Then, these methods were applied to the homemade OBC of <strong>OUFTI</strong>-1. The results<br />

obtained were correlated with ones measured during an experimental test.<br />

Finally, some sensitivity analysis were undertaken to study the eect of several parameters<br />

on the dynamic behavior of the OBC 2.<br />

After this chapter, the following conclusions have to be kept in mind:<br />

• The most accurate method developed in this chapter is the homemade method, which<br />

combines several general methods to create an accurate FE model of the electronic<br />

card. So, it is this method that will be applied in the following chapter.<br />

• The experimental results could be better. So, if an other test is performed next year,<br />

the following facts have to be taken into account:<br />

To come closer to the "free-free" boundary conditions, the card could be xed<br />

by only one hole, using one elastic. The set-up will be more unstable, but the<br />

additional modes that will be excited, are rigid body modes so, there is no<br />

problem.<br />

The hammer excitation is the best solution to excite the structure in "free-free"<br />

boundary conditions. However, it could be better to realize the test using a<br />

stainless steel tip, which will bring a stronger PSD. Note that it will need to<br />

bring attention to the "bounce" eect.<br />

It could be interesting to give up the "free-free" boundary conditions and to<br />

realize a chassis which allows to x the electronic card on a shaker. Thus, the<br />

real boundary conditions of the electronic card's lifetime will be more accurately<br />

reproduced and the results will be better. Note that the chassis have so to be<br />

included in the FE model.<br />

100


Chapter 5<br />

Complete FE analysis of <strong>OUFTI</strong>-1:<br />

Static, modal and random vibration<br />

5.1 Introduction<br />

Through this chapter, a complete FE analysis of <strong>OUFTI</strong>-1 will be performed.<br />

Firstly, the modeling strategy for each of its components will be exposed. A selection<br />

will be realized between these components to choose which of them have a critical inuence<br />

on the global behavior of the entire satellite and so, have to be taken into account in the<br />

FE model, and which can be neglected to limit the modeling eorts.<br />

Then, three dierent analysis will be carried out:<br />

• A static analysis, where the satellite will be submitted to the static loads encountered<br />

during the launch phase.<br />

• A modal analysis, which will allow to obtain the natural frequencies of the satellite<br />

and to verify their fulllment of the ESA requirements.<br />

• A random vibration analysis, where the satellite will be submitted to a random<br />

vibration. This analysis is essential because the decision to accept the satellite on<br />

board Vega (or any other space launcher), is strongly based on its results.<br />

5.2 Modeling strategy<br />

Through this section, the modeling strategy for the relevant components of the satellite<br />

will be exposed. It should be noted that this modeling is performed using the Samcef<br />

software. Firstly, the Catia model presented in chapter 2, is uploaded in the Samcef<br />

101


environment. Then, this model is exploded in several parts to isolate each component.<br />

Finally, a behavior and a material are assigned to each of them and they are linked together<br />

using the assembly methods available in Samcef (the methods used must be as close as<br />

possible from the real connection between the two components to obtain accurate results).<br />

5.2.1 Materials properties<br />

Firstly, the properties of the materials used for manufacturing the several components,<br />

are presented in a single table (see Table 5.1). This allows to avoid multiple repeatings of<br />

tables including these properties through this section. Afterwards, these materials will be<br />

just mentioned and their properties will not be reminded. It should be noted that these<br />

properties were found using the CES software.<br />

Materials Young's moduli (GP a) Poisson ratios Densities (kg/m 3 )<br />

Al − 5052 H32 71.795 0.33675 2685<br />

Al − 5754 69 0.33 2700<br />

Al − 7075 T 6 72.5 0.33 2800<br />

Al − 7075 T 73 71 0.33 2800<br />

FR 4 20 0.3 2200<br />

Stainless steel 205 0.3 7850<br />

Table 5.1: Properties of the materials used in <strong>OUFTI</strong>-1<br />

5.2.2 Main structure<br />

The main structure of <strong>OUFTI</strong>-1, which corresponds to the CSK structural parts,<br />

was modeled using shell components. Indeed, the thickness of the base and end plates<br />

(1.52 mm) and the one of the chassis (1.27 mm) are negligible with regard to their other<br />

dimensions (100 mm), which allows to use this type of modeling. All these components<br />

are made of Al − 5052 H32.<br />

The base and end plates are connected to the chassis by several M3 × 4 mm stainless<br />

steel screws. To model this type of connection in Samcef, the procedure is the following<br />

one:<br />

• A point (vertex) is placed at the center of the two holes (one on the considered plate<br />

and the other on the chassis).<br />

• These vertex are linked to the wire of their respecting hole using a "mean" assembly.<br />

This type of assembly determines the mean rotation and displacement of the nodes<br />

102


concerned. This particularity allows to model the fact that the screw is not rigidly<br />

xed to the structural parts, and that a set can occured. It is important to note that<br />

the denition of this type of assembly must be realized carefully. Indeed, one vertex<br />

is used as the reference point for the determination of the "mean" assembly. It is<br />

important that this point is selected rst and therefore, is dened as support 1. The<br />

vertex must be dened in the "Modeler" module as "not a datum point". The group<br />

of master nodes will be placed on support 2. No other parameters are required [56].<br />

• Finally, the two vertex are linked together using a "xed" assembly, which allows to<br />

model the screw. This type of assembly joins the two parts rigidly together.<br />

The other components, such as the feet (normal and special ones), the deployment<br />

switch, ... were not taken into account because they do not aect signicantly the global<br />

behavior of the satellite.<br />

5.2.3 Solar panels<br />

The solar panels, made of Al − 7075 T 73, were also modeled using shell components<br />

with a thickness of 1.5 mm.<br />

These solar panels will be connected to the dierent faces (+X, +Y , −Y , +Z and −Z)<br />

of the satellite using a specic glue: epoxy Stycast 2850 (with catalyst 24 LV), as alredy<br />

mentioned. To model this type of connection in Samcef, a "glue" assembly is used. This<br />

assembly is currently used when a glue is involved in a connection between two components.<br />

It establishes a linear relation between the nodes on support 1 and the faces of the elements<br />

on support 2 (as shown in Figure 5.1). The mesh is automatically adapted [56].<br />

Figure 5.1: Illustration of a "glue" assembly [56]<br />

The solar cells and the electronic pads were not included in the model for the same<br />

reasons than before (negligible eect).<br />

103


5.2.4 Antenna support<br />

For the antenna support, it was chosen to only model the support itself (made of<br />

Al − 5754). Indeed, due to their negligible mass, the antennas do not play an important<br />

role in the global behavior of the support, as well as the thermal knife and the "Dyneema"<br />

wire. However, owing to its complex geometry, this support can not be modeled using shell<br />

components. So, in this case, volume elements are used (in Samcef, it corresponds to apply<br />

the behavior of "exible volume" on the support).<br />

The xing method is the same as the one used for the solar panels. The support is<br />

connected to the face −X of the chassis using a "glue" assembly.<br />

5.2.5 Electronic cards and battery support<br />

The modeling procedure for the electronic cards was discussed in chapter 4. In this<br />

chapter, it was noted that the best method to realize an accurate FE model of an electronic<br />

card is the homemade method, which uses the following simplications:<br />

• The light components are completely neglected.<br />

• The SMT components are modeled using the global mass smearing method.<br />

• The heavy components, such as the P C 104 connectors, are modeled using a detailed<br />

FE model.<br />

It was also noted that the electronic cards, due to their low thickness (1.7 mm), can be<br />

modeled using shell elements.<br />

So, this method will now be applied to each electronic card individually.<br />

The FM 430 card<br />

This card is presented in Figure 5.2.<br />

Applying the homemade method to this card is not so easy. Indeed, it can be seen that<br />

several heavy components are present on this card. Unfortunately, the properties of these<br />

components are not known so, a detailed FE model of these ones can not be realized. For<br />

this reason, it was decided that the P C 104 connector would be modeled using a detailed<br />

FE model, and the other components would be modeled using the global mass smearing<br />

method. The properties nally obtained for this card, are presented in Table 5.2.<br />

104


Figure 5.2: The FM 430 card from Pumpkin<br />

Components Young's moduli (GP a) Poisson ratios Densities (kg/m 3 )<br />

P C 104 connector 0.846 0.3 1406.923<br />

PCB 20 0.3 3975.473<br />

Table 5.2: Properties of the FM 430 card<br />

The OBC 2 card<br />

This card is presented in Figure 5.3.<br />

Figure 5.3: The OBC 2 card<br />

The application of the homemade method to this particular electronic card, was already<br />

presented in section 4.3.7. The results nally obtained are reminded in Table 5.3.<br />

105


Components Young's moduli (GP a) Poisson ratios Densities (kg/m 3 )<br />

P C 104 connector 0.846 0.3 1406.923<br />

PCB 20 0.3 2822.705<br />

Table 5.3: Properties of the OBC 2 card<br />

The EPS card<br />

This card is presented in Figure 5.4.<br />

Figure 5.4: The EPS card<br />

This electronic card is the one with the greatest number of components. For the most,<br />

these components belong to the SMT category (introduced in section 4.2.5) so, they can be<br />

modeled using the global mass smearing method. The P C 104 connector is still modeled<br />

using a detailed FE approach. The properties nally obtained for this card are presented<br />

in Table 5.4.<br />

Components Young's moduli (GP a) Poisson ratios Densities (kg/m 3 )<br />

P C 104 connector 0.846 0.3 1406.923<br />

PCB 20 0.3 3186.405<br />

Table 5.4: Properties of the EPS card<br />

106


The xEPS card<br />

This card is presented in Figure 5.5.<br />

Figure 5.5: The xEPS card developed in collaboration with Thales Alenia Space ETCA<br />

Except for one (situated at the top of the image, on the left), all the components of<br />

this card belong to the SMT category. So, once again, all the components will be modeled<br />

using the global mass smearing method. The P C 104 connector will still be modeled using<br />

the same method that in the previous cases. The properties nally obtained for this card,<br />

are presented in Table 5.5.<br />

Components Young's moduli (GP a) Poisson ratios Densities (kg/m 3 )<br />

P C 104 connector 0.846 0.3 1406.923<br />

PCB 20 0.3 3404.345<br />

Table 5.5: Properties of the xEPS card<br />

The COM card<br />

Unfortunately, at the moment when this thesis ends, this card is not yet manufactured.<br />

So, no picture of it can be presented. In addition, no accurate information are available<br />

for this card (e.g., concerning its mass). Its nal mass was estimated to maximum 65 g by<br />

the COM subsystem. So, like for the other electronic cards, the P C 104 connector will be<br />

model accurately and the mass of the other components will be spread out over the entire<br />

PCB. The properties nally obtained for this card, are presented in Table 5.6.<br />

107


Components Young's moduli (GP a) Poisson ratios Densities (kg/m 3 )<br />

P C 104 connector 0.846 0.3 1406.923<br />

PCB 20 0.3 4046.952<br />

Table 5.6: Properties of the COM card<br />

Battery support<br />

The battery support is more dicult to model. Indeed, its two parts (the box and the<br />

cover), made of Al − 7075 T 6, have specic geometries which do not allow to model them<br />

using shell elements. So, to be as accurate as possible, it was decided to model these two<br />

parts using volume elements. For the batteries, the problem is quite dierent. These ones<br />

do not have a constant thickness, and their Young's modulus is not known. So, it was<br />

decided to represent them using a local mass smearing method (the mass of the batteries<br />

can be obtained by weighting them or by consulting their datasheet). As remind, the<br />

density of the battery is 2159.25 kg/m 3 . Then, the batteries are xed to their support<br />

using a "glue" assembly.<br />

Assembly<br />

These ve electronic cards and the battery support are all xed on the endless screws.<br />

To realize these connections, some vertex were added to each endless screws, at the level<br />

of each of these components. These vertex are used to create the endless screws and so,<br />

are rigidly xed to them. Then, a "mean" assembly is applied between each vertex and<br />

the wire of its corresponding hole on the considered component.<br />

5.2.6 The endless screws<br />

The endless screws, due to their higher dimensions and to the fact that they are involved<br />

in a lot of connections (as explained in section 5.2.5), will be modeled more accurately than<br />

the screws which connect the CSK structural parts together. These screws are M3 × 8 cm<br />

stainless steel screws. So, to model them accurately, a circular beam behavior was used,<br />

with a diameter of 2.459 mm, which corresponds to the M3 standard.<br />

In reality, these screws are connected to the base plate at one extremity, and to the<br />

chassis at the other, thanks to the midplane standos. For the rst connection, the same<br />

assembly method as the one used for the electronic cards and the battery support, is<br />

applied. For the second one, an homemade method was imagined. The midplane standos<br />

were replaced by two vertex connected together using a "xed" assembly. To clarify the<br />

following explication, Figure 5.6 shows this particular modeling. One of these vertex (1) is<br />

108


located at the extremity of the considered endless screw, and the other (2) is situated at<br />

the same level, but in front of the hole of the chassis which allows to connect this one with<br />

the considered midplane stando. Then, the screw which realizes the connection between<br />

the chassis and the considered midplane stando, is modeled using the same method that<br />

for the connection between the base and end plates, and the chassis (by using two vertex<br />

(2 and 3), xed rigidly one to the other).<br />

Figure 5.6: Modeling of the midplane standos<br />

5.2.7 ADCS and THER components<br />

The ADCS components, including the permanent magnet and the four hysteretic bars,<br />

are not modeled. The same decision was taken for the THER components, including<br />

heaters, thermostats and thermal insulation. This choice was made for several reasons:<br />

• First, due to their restricted dimensions and mass, they do not play a signicant role<br />

in the global behavior of the satellite.<br />

• In addition, they are all made of materials which present special properties so, these<br />

ones are dicult to obtain.<br />

• Finally, even if these properties can be obtained, they will be modify during the<br />

manufacturing of these elements (in the case of the permanent magnet and hysteretic<br />

bars) and so, the results obtained with and without these components, will not be<br />

more accurate one compared to the other.<br />

Now that the modeling of each part of the satellite is dened, the analysis can start.<br />

109


5.3 Static analysis<br />

As described in reference [57], Vega, and so, its payloads, is subjected to quasi-static<br />

loads during ight. So, the aim of this study is to ensure the structural integrity of each<br />

part of the satellite when it is submitted to these steady state acceleration.<br />

5.3.1 Loads denition<br />

The rst step is to clearly dene the quasi-static loads which will be applied on our<br />

CubeSat. For this, the worst case will be considered. This particular case occures if<br />

<strong>OUFTI</strong>-1 is placed in third position inside the P-POD because, in this conguration, it<br />

has to support the weight of two CubeSats in addition of the quasi-static loads imposed<br />

by the launcher. This conguration is illustrated in Figure 5.7. As remind, the P-POD<br />

will be placed at an angle of 10 ◦ from the vertical.<br />

Figure 5.7: Illustration of the worst case conguration<br />

The maximal steady state acceleration given in reference [57], is the one along the<br />

launcher's longitudinal axis. This acceleration has a value of 6.3 g. In our case, it was<br />

decided to apply a SF of 2 on each acceleration. So, the acceleration along the Z axis of<br />

the Vega frame is 12.6 g. The lateral accelerations, due to the dynamic pressure inside the<br />

launcher, have a maximal value of 1.2 g. So, an acceleration of 2.4 g will be applied along<br />

X and Y axes of the Vega frame.<br />

However, due to the particular conguration of the P-POD, a rotation matrix must be<br />

applied to these accelerations to bring them back in the CubeSat frame (as illustrated in<br />

Figure 5.8).<br />

Note that the CubeSat are placed with their base plate (face −Z) on the side +Z inside<br />

the P-POD (as it can be seen in Figure 5.8). This conguration is the one given in the<br />

ocial documents [58].<br />

110


Figure 5.8: Frame rotation<br />

The calculation of the steady state accelerations in the CubeSat frame is presented in<br />

Equation 5.3.1.<br />

⎛ ⎞ ⎛<br />

⎞ ⎛ ⎞ ⎛ ⎞<br />

a X, CubeSat 1 0 0<br />

2.4 2.4<br />

⎝a Y, CubeSat<br />

⎠ = ⎝0 cos (10 ◦ ) −sin (10 ◦ ) ⎠ × ⎝ 2.4 ⎠ = ⎝ −0.1756 ⎠ (5.3.1)<br />

a Z, CubeSat 0 sin (10 ◦ ) cos (10 ◦ ) 12.6 −12.8253<br />

Owing to the particular case which is considered, the weight of the two other CubeSats<br />

placed inside the P-POD must be added to these acceleration. The maximal acceptable<br />

mass for one CubeSat is 1 kg. So, the forces acting on the four feet of <strong>OUFTI</strong>-1 base plate,<br />

due to the two CubeSats placed above it, are calculated in Equation 5.3.2.<br />

F = m × g × a quasi−static = 2 × 9.81 × 12.8253 = 251.64 N ≈ 252 N (5.3.2)<br />

which lead to a force of 63 N on each foot along the Z axis of the CubeSat frame.<br />

It is important to note that the accelerations considered in this analysis occured at<br />

dierent phases of the ight (the maximal longitudinal acceleration is reached at the third<br />

stage's maximal acceleration and maximal lateral accelerations are reached during the lifto<br />

and under the ight maximal dynamic pressure). So, the resulting model is conservative,<br />

which brings an additional security with regard to the results.<br />

An illustration of the complete loading state (in the CubeSat frame) is given in Figure<br />

5.9.<br />

111


Figure 5.9: Illustration of the complete quasi-static loading state of <strong>OUFTI</strong>-1<br />

5.3.2 Boundary conditions<br />

To determine the boundary conditions to apply to our CubeSat, its mounting within<br />

the P-POD must be known. This one is available in reference [58]. Then, these conditions<br />

must be dened as follow:<br />

• The CubeSat is placed inside the P-POD with its base plate on the +Z side. So,<br />

in the worst case considered here, the end plate's feet are in direct contact with the<br />

P-POD. For this reason, these feet will be clamped to simulate this perfectly rigid<br />

contact.<br />

• In addition, the rails situated in the corners of the CubeSat are in contact with the<br />

rails of the P-POD. So, some lockings (along X and Y axes) will be applied on the<br />

rails of the CubeSat to represent this linear contact.<br />

5.3.3 Results<br />

The results nally obtained are presented in Table 5.7 and in Figures 5.10 and 5.11.<br />

Maximal displacement (mm) Maximal Von Mises stress (MP a)<br />

0.0909 101.97<br />

Table 5.7: Results of the static analysis<br />

112


Figure 5.10: Displacements of <strong>OUFTI</strong>-1<br />

In Figure 5.10, it can be seen that the maximal displacements of <strong>OUFTI</strong>-1 are along<br />

the Z axis, which is the longitudinal direction of the P-POD. More particularly, it is the<br />

center of electronic cards which presents the greatest displacement. This is in complete<br />

accordance with the loading state. Indeed, the structure supports the greatest acceleration<br />

along this particular axis.<br />

In addition, it can be noted that the base and end plates present some deformations<br />

at the feet location, which are due to the boundary conditions and to the loads applied to<br />

them.<br />

In Figure 5.11, it can be seen that the maximal Von Mises stress appears around the feet<br />

of the end plate. This is, once again, in accordance with the conditions applied. Indeed,<br />

these feet are clamped, which is a the strongest restrictive constraint.<br />

113


Figure 5.11: Maximal Von Mises stress inside <strong>OUFTI</strong>-1<br />

According to the reference [59], it is possible to dene a MoS for the entire satellite.<br />

This calculation is performed in Equation 5.3.3.<br />

where:<br />

MoS =<br />

allowable load<br />

(applied load) − SF − 1 (5.3.3)<br />

• allowable load: Allowable load under specied functional conditions (e.g., yield,<br />

buckling, ultimate)<br />

• applied load: Computed or measured load under dened load condition (design loads)<br />

• SF : Safety factor applicable to the specied functional conditions including the<br />

specied load conditions (e.g., yield, buckling, ultimate)<br />

For <strong>OUFTI</strong>-1, the maximal Von Mises stress appears in the end plate, which is made<br />

of Al − 5052 H32. This material has a yield stress of 162 MP a, the maximal applied stress<br />

is 101.97 MP a and the SF used for this study is 2. So, the value nally obtained for the<br />

MoS is 0.5887. It means that <strong>OUFTI</strong>-1 can support a loading case 58.87% higher than<br />

the one used during this analysis. It also means that our CubeSat fullls well the ESA<br />

requirement, which stipulates that all MoS must be positive.<br />

This last result marks the end of our static analysis. Now, a modal analysis will be<br />

performed to estimate the global dynamic response of <strong>OUFTI</strong>-1.<br />

114


5.4 Modal analysis<br />

This analysis will allow to predict the global dynamic behavior of <strong>OUFTI</strong>-1.<br />

In this analysis, the CubeSat is modeled using exactly the same strategy as in section<br />

5.2. Only the static loads and the boundary conditions are removed to obtain the behavior<br />

of <strong>OUFTI</strong>-1 in "free-free" boundary conditions. This type of study allows to obtain a rst<br />

idea of its behavior in real life conditions in a simple way. In addition, the results can<br />

be used to realize a model updating. Indeed, this conguration has the advantage that<br />

it does not introduce additional unknowns about the boundary conditions. So, the model<br />

updating can be performed more accurately. However, these conditions are really dicult<br />

to simulate during an experimental test and some approximations must be done (but it is<br />

not the topic that we want to develop here).<br />

5.4.1 Results<br />

It was decided to compute the ve rst modes of <strong>OUFTI</strong>-1. Their natural frequencies<br />

are listed in Table 5.8.<br />

Modes (mm) Natural frequencies (Hz)<br />

1 325.51<br />

2 333.92<br />

3 343.27<br />

4 368.63<br />

5 377.29<br />

Table 5.8: Five rst natural frequencies of <strong>OUFTI</strong>-1 in hard-mounted conguration<br />

It can be noted that all these frequencies turn around a value of 350 Hz. This is due to<br />

the fact that each of these modes excites only the electronic cards of <strong>OUFTI</strong>-1 (and their<br />

xations).<br />

These results prove that <strong>OUFTI</strong>-1 respects the ESA requirement on the natural frequencies:<br />

"The Cubesats shall not have structural modes at frequencies lower than 120 Hz<br />

in hard-mounted conguration".<br />

Now, the rst two modes will be described (the following ones present the same behavior<br />

and so, it does not have a great interest to analyze them):<br />

• The rst mode of <strong>OUFTI</strong>-1, which corresponds to a natural frequency of 325.51 Hz,<br />

is presented in Figure 5.12.<br />

115


Figure 5.12: First mode of <strong>OUFTI</strong>-1 (325.51 Hz)<br />

This mode is characterized by a vibration, in phase opposition, of the two extreme<br />

electronic cards (the FM 430 on the side −Z and the COM on the side +Z). The<br />

other electronic cards, the main structure and the external panels are not excited by<br />

this mode.<br />

• The second mode of <strong>OUFTI</strong>-1, which corresponds to a natural frequency of 333.92 Hz,<br />

is presented in Figure 5.13.<br />

Figure 5.13: Second mode of <strong>OUFTI</strong>-1 (333.92 Hz)<br />

This mode is characterized by a vibration of most electronic cards. Indeed, the FM<br />

430 and the COM present oscillations in phase, just like the EPS and xEPS. These<br />

two groups of cards vibrate in phase opposition one from each other. It can also be<br />

seen that the endless screws are more involved in this mode than in the rst one.<br />

The OBC 2 card, the main structure and the external panels are not excited by this<br />

mode.<br />

116


Next year, several tests will have to be performed to validate this model and to verify<br />

the exact fulllment of the ESA requirements.<br />

Now, the last analysis which must be performed on <strong>OUFTI</strong>-1, concerns random vibrations.<br />

5.5 Random vibration<br />

This analysis is based on the characteristics of the LV which is selected. Indeed, this<br />

study needs several information about the vibration environment that the satellite has to<br />

withstand during launch, and these information are specic to each launcher.<br />

For CubeSats, this analysis is generally limited to random vibration. These vibrations<br />

are generated by the propulsion system of the launcher and by the vibration-acoustic<br />

response of adjacent structures. Maximum excitation levels occured during the principal<br />

stage ignition. To characterize these vibrations, each launcher possesses its own PSD. The<br />

typical PSDs of several LVs, at qualication level, are given in Figure 5.14. The last version<br />

of the Vega PSD is given in Figure 5.15.<br />

Figure 5.14: PSD of several LVs at qualication level [60]<br />

117


Figure 5.15: Precise PSD of Vega [57]<br />

The model used for this analysis is the same as the one used to perform the modal<br />

analysis. To be more accurate, this model would have to integrate the P-POD FE model,<br />

to represent accurately the boundary conditions really imposed to the CubeSat during the<br />

launch phase.<br />

To realize a random vibration analysis, the FE model of the satellite must be excited<br />

at the level imposed by the selected launcher. In the case of Vega, for qualication level<br />

(which is the most restrictive level), the PSD can be decomposed as follow:<br />

• 0.07 g 2 /Hz in the frequency range: [20 − 60 Hz]<br />

• From 0.07 g 2 /Hz to 0.1 g 2 /Hz in the frequency range: [60 − 70 Hz]<br />

• 0.1 g 2 /Hz in the frequency range: [70 − 200 Hz]<br />

• From 0.1 g 2 /Hz to 0.2 g 2 /Hz in the frequency range: [200 − 300 Hz]<br />

• 0.2 g 2 /Hz in the frequency range: [200 − 600 Hz]<br />

• From 0.2 g 2 /Hz to 0.02 g 2 /Hz in the frequency range: [600 − 2000 Hz]<br />

This particular PSD must be introduced in Samcef software through a ".psd" le.<br />

Unfortunately, due to a problem encountered with Samcef (which is not yet resolved),<br />

this analysis can not be performed this year. However, all the necessary les were created,<br />

and information were collected to facilitate the work of the next year's student.<br />

118


5.6 Summary<br />

Through this chapter, a complete FE analysis of <strong>OUFTI</strong>-1 was performed.<br />

First, the modeling strategy used to represent each component of the satellite in the<br />

FE model, was described accurately.<br />

Then, a static analysis was performed. This analysis allows to conclude that <strong>OUFTI</strong>-1<br />

withstands the quasi-static loads imposed by Vega in the worst case conguration. So, the<br />

structural integrity of the satellite in static state is ensured.<br />

The modal analysis led to results which satisfy ESA requirements on the dynamic behavior<br />

of the Vega payloads, and showed that the most critical parts of the satellite are<br />

its electronic cards, which are strongly excited in the rst modes. Next year, several tests<br />

will have to be performed to validate this model.<br />

Finally, the procedure to realize a random vibration analysis was drawn. Each step of<br />

this analysis was presented and resolved. However, due to an unsolved problem encountered<br />

with the Samcef software, this analysis could not be realize for this thesis. Once this<br />

analysis will be performed and when a complete engineering model of the satellite will be<br />

availble, qualication tests will have to be performed (strictly following the ESA procedure<br />

[61]) to denitely validate the satellite and to send them to Kourou for its launch.<br />

119


Chapter 6<br />

Conclusions<br />

This master thesis developed the structural design and dynamic analysis of <strong>OUFTI</strong>-1.<br />

The main objectives were to design a new reliable support for the batteries, which fullls<br />

all the requirements imposed by several subsystems, and to create an accurate FE dynamic<br />

modeling procedure for the electronic cards.<br />

Through this chapter, the results obtained during all the year, will be summarized.<br />

Then, perspectives for future works will be highlighted.<br />

Finally, the parallel activities linked to the project, will be briey exposed.<br />

6.1 Summary of the accomplished work<br />

The rst chapter of this thesis consisted in a general presentation of the project. The<br />

CubeSat concept was introduced to allow to the profane to understand what is a CubeSat<br />

and what are the principal requirements and restrictions which are imposed to this particular<br />

type of satellite. Then, the genesis of the <strong>OUFTI</strong>-1 project was presented to expose the<br />

guideline of the project from its starting point to the level at which this thesis started. The<br />

launch opportunity oered by the ESA, was also discussed. Finally, the mission payloads<br />

and the <strong>OUFTI</strong> team were described to nalize this overall view of the <strong>OUFTI</strong>-1 project.<br />

Chapter 2 presented the general ight system conguration. The reference frame used to<br />

locate precisely each component of the CubeSat, was dened. Then, each part of <strong>OUFTI</strong>-1<br />

was presented in a global understanding approach. The technical constraints which led to<br />

this particular organization, were highlighted, and the methods which will be used to x<br />

the dierent components together, were exposed. Finally, general physical and mechanical<br />

properties, including: the CoG location, the inertia properties and the mass budget, were<br />

120


calculated, and their fulllment of all requirements imposed by ESA, was veried.<br />

Chapter 3 developed the complete battery support design, from the initial idea to the<br />

manufacturing of this support. The reasons that led to modify the last year's design, were<br />

also presented. This design included:<br />

• The selection of the several components which had to be incorporated inside the<br />

support, including: batteries, thermal control's devices such as thermostats, heaters<br />

and thermal insulator, was performed.<br />

• Then, the material which was used to manufacture the support, was selected. This<br />

material has to fulll several physical, mechanical and thermal requirements. This<br />

step is really important because the thickness of the dierent parts that must prevent<br />

the batteries' bulge, was determined in relation to the properties of the selected<br />

material.<br />

• Once the material and the components were selected, the design phase was presented.<br />

The procedure was introduced step by step, just like it progressed during the whole<br />

year.<br />

• Finally, the design was validated by several calculations, such as the mass budget,<br />

and FE analysis, including static and dynamic analysis.<br />

Chapter 4 exposed a general procedure to create an accurate FE dynamic model of<br />

electronic cards. Firstly, several existing methods were presented and their accuracy was<br />

studied. Then, these methods were applied to the particular case of the homemade OBC<br />

of <strong>OUFTI</strong>-1. A homemade method, based on the fact that several methods can be applied<br />

to particular categories of components soldered on a same PCB, was also created. This<br />

method is the one which presented the best results and so, it is the one that was applied to<br />

each electronic card of <strong>OUFTI</strong>-1. Finally, some sensitivity analysis with regard to several<br />

parameters which strongly inuence the PCB dynamic response, were carried out.<br />

Finally, chapter 5 consisted in a complete FE analysis of the satellite. Firstly, the<br />

modeling strategy used to represent each component of the satellite in the FE model, was<br />

described accurately. Then, a static analysis was performed, and the results obtained ensure<br />

that <strong>OUFTI</strong>-1 withstands well the quasi-static loads which will be applied to it during<br />

the launch phase. The modal analysis led also to results that fulll all ESA requirements.<br />

Finally, the procedure to realize a random vibration analysis was presented and all necessary<br />

les were created.<br />

Throughout this thesis, we have tried to develop a detailed and rigorous description of<br />

the approach followed during all our progression. To realize this, a theoretical basis was<br />

121


provided for each method used. We have also tried to express clearly the several reasons<br />

which led us to use these particular methods, as well as the assumptions to respect for<br />

applying them, because we see in these assumptions one of the cornerstones for the success<br />

of a rigorous scientic work.<br />

6.2 Perspectives<br />

This section intends to give some guidelines for the future student who will be in charge<br />

of the structural subsystem. In the summary of each chapter, the problems encountered<br />

were highlighted. This section is here to remind these problems and to highlight them once<br />

again.<br />

Concerning the general ight system conguration, most of the problems were solved.<br />

The order of the electronic cards was conrmed and xed. The radiance diagram of the antennas<br />

was realized by the RF subsystem, which allowed to denitely conrm the location<br />

of the permanent magnet (in accordance with the requirements of the ADCS subsystem).<br />

This allowed also to dened the location of the four hystereric bars inside the CubeSat.<br />

The next problem will be to realize a complete integration procedure for the ight model<br />

of <strong>OUFTI</strong>-1. A basic procedure for the integration of the engineering model was already<br />

composed this year [33], but it must be completed by several information (how to include<br />

the cables inside the satellite, how to connect the several subsystems together, ...). The<br />

mass budget must also be denitely conrmed by weighting the components which still<br />

present some uncertainties about their mass.<br />

Concerning the battery support, the manufacturing has to be nalized. Then, thermal<br />

and structural tests must be performed to be correlated with the models created this year.<br />

This step will allow to denitely validate this design and to ensure its structural integrity<br />

during all the CubeSat lifetime.<br />

Concerning the electronic cards, an other experimental test will have to be performed,<br />

following advice given in chapter 4, to denitely validate the fact that the homemade<br />

method is the one which gives the best results. The physical properties of all components,<br />

such as the Young's modulus, can also be determined to allow to model them using global<br />

or local smearing methods which include the stiness contributions of these components.<br />

Finally, the experimental modal survey of <strong>OUFTI</strong>-1 in its hard-mounted version, will<br />

have to be performed and correlated to the FE model presented in the last chapter of this<br />

thesis. In addition, the random vibration analysis and tests for the launch phase must also<br />

be performed to qualify <strong>OUFTI</strong>-1 for its travel to space.<br />

122


6.3 Parallel activities<br />

This project, which is a particular subject for a master thesis, gives to the students the<br />

opportunity to take part in a lot of parallel activities, such as ESA conferences or CubeSat<br />

meetings all over the world. So, the activities, in which I am directly involved, performed<br />

during this academic year, will be presented here:<br />

• The rst activity, which set at Royal Meteorological Institute (IRM) based at Bruxelles<br />

(Belgium), was a general public presentation. During this activity, the goal<br />

was to present the project to people that has no knowledge, or nearly, of the space<br />

activities. It allows us to go one's rst steps inside the project and to familiarize us<br />

with its multidisciplinary aspects.<br />

• The second activity, which set at CSL (Belgium), was a conference with the Del-C 3<br />

team. During this activity, presentations of the two teams' projects were realized. It<br />

allows us to exchange several points of view with them and to move forward in our<br />

understanding of constraints imposed by the space environment.<br />

• The following activity was the Interface Technical Review (ITR), which set at the<br />

EuroSpace Center of Redu (Belgium). At this moment, each student had all the technical<br />

background necessary to accomplish his personal work. Before going to Redu,<br />

each of them had to provide a data package including all the interfaces between his<br />

subsystem and the other ones, and internal interfaces inside his own subsystem [62].<br />

For each of these interfaces, a technical solution was proposed by the students who<br />

are in charge of the concerned subsystems. Then, during the weekend spent to Redu,<br />

several meetings were realized. These meetings concerned the four general categories<br />

of interfaces that exist in a satellite, including: mechanical, thermal, electrical and<br />

data interfaces. They involved the students of the <strong>OUFTI</strong> team, the system engineering<br />

team, the project managers, the professors and several industrial specialists.<br />

During these meetings, each solution was studied with the help of all this sta, and<br />

was approved or improved to fulll the requirements imposed to it. This activity allowed<br />

to dene precisely all the interfaces present inside the CubeSat and to highlight<br />

them, and brought to each student involved in the project, a better understanding<br />

of the overall satellite. It allowed also to take a big step for the project achievement<br />

by emerging from an Interface Control Document (ICD) for <strong>OUFTI</strong>-1.<br />

• Finally, the last activity during this year, was the Workshop on the Verication of<br />

the CubeSats for the Vega Maiden Flight, which set at ESTEC (The Netherlands).<br />

During this workshop, the verication concept and the acceptance data package were<br />

introduced. These concepts are really important because they include all the verications<br />

which must be performed on each CubeSat before delivering it to ESA. The<br />

123


last versions of the Verication Control Document (VCD) and the ICD were also introduced.<br />

These documents were covered point by point to explain to the CubeSats<br />

teams what ESA expects from them for the qualication of their CubeSats. Several<br />

presentations on the structural verication and collision & contamination analysis<br />

were also performed by ESA experts. Finally, each team had defended its project<br />

through a general presentation about their schedule for the following months. This<br />

activity allows us to meet people coming from all over the world, and to take advantage<br />

of their own experience to learn more and more about this fascinating world<br />

that the space is.<br />

124


Bibliography<br />

[1] S. LEE et al. "CubeSat Design Specication, Revision 12 ". California Polytechnic<br />

State University, San Luis Obispo, United States, 2009.<br />

[2] "Pumpkin Website". http://www.cubesatkit.com.<br />

[3] V. BASHBUSH. "Characterization of the Internal and External Environments of the<br />

CubeSat P-POD and Test Pod ". <strong>Master</strong> <strong>Thesis</strong>, California Polytechnic State University,<br />

San Luis Obispo, United States, 2004.<br />

[4] S. GALLI. "Mission Design for the CubeSat <strong>OUFTI</strong>-1 ". <strong>Master</strong> <strong>Thesis</strong>, University<br />

of Liège, Belgium, 2008.<br />

[5] J. PISANE. "Design and Implementation of the Terrestrial and Space Telecommunication<br />

Elements of the Student Nanosatellite of the University of Liège ". <strong>Master</strong> <strong>Thesis</strong>,<br />

University of Liège, Belgium, 2008.<br />

[6] V. BEUKELAERS. "From Mission Analysis to Space Flight Simulation of the<br />

<strong>OUFTI</strong>-1 Nanosatellite". <strong>Master</strong> <strong>Thesis</strong>, University of Liège, Belgium, 2009.<br />

[7] L. CHIARELLO. "Design and Implementation of the Control and Monitoring System<br />

for the Ground Station of Nanosatellite <strong>OUFTI</strong>-1 of University of Liège ". <strong>Master</strong><br />

<strong>Thesis</strong>, University of Liège, Belgium, 2009.<br />

[8] N. EVRARD. "Développement de l'Infrastructure de Mesure du Nanosatellite <strong>OUFTI</strong>-<br />

1 ". <strong>Master</strong> <strong>Thesis</strong>, Gramme Institute, Belgium, 2009.<br />

[9] S. HANNAY. "Modeling of the Attitude Control of the Nanosatellite <strong>OUFTI</strong>-1 ".<br />

<strong>Master</strong> <strong>Thesis</strong>, University of Liège, Belgium, 2009.<br />

[10] J. HARDY. "Implémentation du Protocole AX.25 à Bord du Nanosatellite <strong>OUFTI</strong>-<br />

1 ". <strong>Master</strong> <strong>Thesis</strong>, Gramme Institute, Belgium, 2009.<br />

[11] R. HENRARD. "Réalisation du Système de Télécommunication du Satellite <strong>OUFTI</strong>-<br />

1 ". <strong>Master</strong> <strong>Thesis</strong>, ISIL Institute, Belgium, 2009.<br />

[12] L. JACQUES. "Thermal Design of the <strong>OUFTI</strong>-1 Nanosatellite ". <strong>Master</strong> <strong>Thesis</strong>,<br />

University of Liège, Belgium, 2009.<br />

125


[13] Ph. LEDENT. "Design and Implementation of On-Board Digitally Controlled Electrical<br />

Power Supply of Student Nanosatellite <strong>OUFTI</strong>-1 of University of Liège ". <strong>Master</strong><br />

<strong>Thesis</strong>, University of Liège, Belgium, 2009.<br />

[14] F. MAHY. "Design and Implementation of On-Board Telecommunication System<br />

of Student Nanosatellite <strong>OUFTI</strong>-1 of University of Liège ". <strong>Master</strong> <strong>Thesis</strong>, University of<br />

Liège, Belgium, 2009.<br />

[15] G.PIERLOT. "<strong>OUFTI</strong>-1: Flight System Conguration and Structural Analysis ".<br />

<strong>Master</strong> <strong>Thesis</strong>, University of Liège, Belgium, 2009.<br />

[16] D. TENEY. "Design and Implementation of On-Board Processor and Software of<br />

Student Nanosatellite <strong>OUFTI</strong>-1 ". <strong>Master</strong> <strong>Thesis</strong>, University of Liège, Belgium, 2009.<br />

[17] P. THIRION. "Design and Implementation of On-Board Electrical Power Supply<br />

of Student Nanosatellite <strong>OUFTI</strong>-1 of University of Liège ". <strong>Master</strong> <strong>Thesis</strong>, University of<br />

Liège, Belgium, 2009.<br />

[18] J. WERTZ. "Conception et Réalisation du Système de Déploiement des Antennes<br />

du Nanosatellite <strong>OUFTI</strong>-1 ". <strong>Master</strong> <strong>Thesis</strong>, Gramme Institute, Belgium, 2009.<br />

[19] C. BERNA & M.-A. LURON. "Vega User's Manual, Issue 3, Revision 0 ". Arianespace,<br />

Boulevard de l'Europe, B.P. 177, 91006 Evry - Courcouronnes cedex, France,<br />

2006.<br />

[20] P. GALEONE. "LARES System Pictures". Workshop on the Verication of the<br />

CubeSats for the Vega Maiden Flight, ESA-ESTEC, Noordwijk, The Netherlands, 2010.<br />

[21] "<strong>OUFTI</strong>-1 Website". http://www.leodium.ulg.ac.be/cmsms/index.phppage=satellite.<br />

[22] A. KALMAN. "CubeSat Kit User Manual ". Pumpkin Inc., San Francisco, United<br />

States, 2005.<br />

[23] A. KALMAN. "Solar Panel Clips Set". Pumpkin Inc., San Francisco, United<br />

States, 2007.<br />

[24] E. BOSCH. "Detailed Design and Analysis of the Antenna Deployment System of<br />

the Satellite <strong>OUFTI</strong>-1 ". <strong>Master</strong> thesis, University of Liège, Belgium, 2010.<br />

[25] V. FRANCOIS. "Study of Passive and Active Attitude Control Systems for the<br />

<strong>OUFTI</strong> Nanosatellites". <strong>Master</strong> thesis, University of Liège, Belgium, 2010.<br />

126


[26] P. HARIS. "PC/104-Plus Specication, Version 2.2 ". PC/104 Embedded Consortium,<br />

San Francisco, United States, 2007.<br />

[27] K. CAN. "Projet de Nanosatellite <strong>OUFTI</strong>-1: Développement du Logiciel Embarqué".<br />

<strong>Master</strong> thesis, University of Liège, Belgium, 2010.<br />

[28] N. CROSSET. "Implémentation du Relais D-STAR à Bord du Nanosatellite <strong>OUFTI</strong>-<br />

1 ". <strong>Master</strong> thesis, GRAMME Institute, Belgium, 2010.<br />

[29] N. MARCHAL. "Design, Implémentation et Test de l'Architecture du Système de<br />

Télécommunication du CubeSat <strong>OUFTI</strong>-1 ". <strong>Master</strong> thesis, GRAMME Institute, Belgium,<br />

2010.<br />

[30] A. KALMAN. "Midplane Stando Kit ". Pumpkin Inc., San Francisco, United<br />

States, 2008.<br />

[31] T. KOIZUMI et al. "Identication of the Center-of-Gravity and Inertia Terms of<br />

3-Dimensional Body". Department of Engineering, Doshisha University, Japan, 2006.<br />

[32] "Dassault Systemes - Catia Website ". http://www.3ds.com/products/catia/welcome/.<br />

[33] N. FRANCOIS. "Integration Procedure for the Engineering Model of <strong>OUFTI</strong>-1 ".<br />

Technical Report, University of Liège, Belgium, 2010.<br />

[34] "Samtech Website". http://www.samcef.com/page.aspid=3&langue=EN<br />

[35] J.-Ph. NOEL. "Thermal Issues Settlement and Test Procedure Investigation of<br />

<strong>OUFTI</strong>-1 Nanosatellite". <strong>Master</strong> thesis, University of Liège, Belgium, 2010.<br />

[36] "Kokam Catalog". http://kokamamerica.com/kokam_catalog.pdf.<br />

[37] J. PEDERSEN. "Vega's CubeSats Design Specication ". Workshop on the Verication<br />

of the CubeSats for the Vega Maiden Flight, ESA-ESTEC, Noordwijk, The Netherlands,<br />

2010.<br />

[38] M. ASHBY. "Materials Selection in Mechanical Design ". Butterworth-Heinemann,<br />

1999.<br />

[39] J. LECOMTE-BECKERS. "Materials Selection". LTAS, University of Liège, Belgium,<br />

2009.<br />

127


[40] "Dejond - Aluminum Catalog ".<br />

http://www.dejond.com/EN/Downloads/metals/Dejond%20aluminium.pdf.<br />

[41] "Minco - Polyimide Thermofoil Heaters ".<br />

http://www.minco.com/uploadedFiles/Products/Thermofoil_Heaters/Kapton_Heaters/<br />

hs202b-hk.pdf.<br />

[42] "Klixon 3BT and 4BT Series Catalog ".<br />

[43] "Red Book - Aerospace Thermal Control Materials ".<br />

http://www.sheldahl.com/Product/bulletins/Redbook.pdf, 2007.<br />

[44] R. AMY, G. AGLIETTI & G. RICHARDSON. "Accuracy of Simplied Printed<br />

Circuit Board Finite Element Models ". Microelectronics Reliability 50 (2010), 86-97.<br />

[45] R. AMY, G. AGLIETTI & G. RICHARDSON. "Sensitivity Analysis of Simpli-<br />

ed Printed Circuit Board Finite Element Models ". Microelectronics Reliability 49 (2009),<br />

791-799.<br />

[46] D. WATTIAUX, O.VERLINDEN & Ch. DE FRUYTIER. "Modelling of the Dynamic<br />

Behaviour of Modular Equipments Designed by Thales Alenia Space ETCA".<br />

[47] R. AMY, G. AGLIETTI & G. RICHARDSON. "Simplied Modelling of Printed<br />

Circuit Boards for Spacecraft Applications ". Acta Astronautica 65 (2009), 192-201.<br />

[48] R. AMY, G. AGLIETTI & G. RICHARDSON. "Analysis of Simplied FE Models<br />

of PCBs Exposed to Random Vibrations". Proceedings of ISMA 2008: Durability Testing<br />

- Vibration Control, 979-986.<br />

[49] J. PITARRESI et al. "Dynamic Modeling and Measurement of Personal Computer<br />

Motherboards". Electronic Components and Technology Conference (2002), 597-603.<br />

[50] D. BAKER & Y. CHEN. "Modeling the Vibration Restraints of Wedge Lock Card<br />

Guides". ASME, Journal of Electronic Packaging 115 (1993), Issue 2, 189-194.<br />

[51] D.J. EWINS. Modal Testing: Theory, Practice and Application. Second Edition,<br />

Research Studies Press LTD, 2000.<br />

[52] N. MAIA & J. SILVA. Theoretical and Experimental Modal Analysis. Research<br />

Studies Press LTD, 1997.<br />

128


[53] J.-C. GOLINVAL. "Vibration Testing and Experimental Modal Analysis ". LTAS,<br />

University of Liège, Belgium, 2010.<br />

[54] D. STEINBERG. "Vibration Analysis for Electronic Equipments ". John WILEY<br />

& Sons, 2000.<br />

[55] "Deltatec Website". http://www.deltatec.be/<br />

[56] "Samcef-Field Help".<br />

[57] J. PEDERSEN. "Vega CubeSats ICD, Issue 0, Revision 4 ". ESA-ESTEC, Noordwijk,<br />

The Netherlands, 2009.<br />

[58] J. BROWN. "Test Pod User's Guide Revision 6 ". California Polytechnic State<br />

University, San Luis Obispo, United States, 2006.<br />

[59] "ECSS-E-30 Part 2A - Space Engineering - Mechanical - Part 2: Structural ".<br />

ESA-ESTEC, Requirements & Standards Division, Noordwijk, The Netherlands, 2000.<br />

[60] G. ROETHLISBERGER. "SwissCube - Phase C: Launch Environment ". Lausanne,<br />

Switzerland, 2008.<br />

[61] P. GALEONE. "Approach for the Structural Tests for the Qualication of the<br />

CubeSats/P-PODs". ESA-ESTEC, Noordwijk, The Netherlands, 2009.<br />

[62] N. FRANCOIS. "ITR - Data Package: Subsystem STRU ". Technical Report,<br />

University of Liège, Belgium, 2010.<br />

129


Appendix A<br />

Schemas of solar panels<br />

Figure 6.1: Schemas of solar panels +X and +/ − Y [15]<br />

130


Figure 6.2: Schemas of the solar panel +Z [15]<br />

Figure 6.3: Schemas of the solar panel −Z [15]<br />

131


Appendix B<br />

Schemas of electronic cards and their available volume<br />

Figure 6.4: Schemas of electronic cards [15]<br />

132


Figure 6.5: Available volume for each electronic card [62]<br />

133


Appendix C<br />

Fixations of the endless screws<br />

Figure 6.6: Fixations of endless screws on the base plate [62]<br />

134


Figure 6.7: Picture of the xation between one midplane stando and the chassis [33]<br />

Figure 6.8: Disposition of midplane standos inside the CubeSat [30]<br />

135


Appendix D<br />

Datasheet of the battery Kokam SLB 603870H<br />

136


Figure 6.9: Datasheet of the battery Kokam SLB 603870H<br />

137

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!