1. Challenge the future Delft University of Technology Small Satellite Reliability Research on Spacecraft Under 50 Kg: Analysis on Component Level Jan Kolmas, Jian Guo, Eberhard Gill

2. 2 Outline Introduction Small Satellites Anomalies Database 50 Non-parametric Analysis Parametric Analysis Case Study Conclusions

3. 3 Introduction Research Motivation Increasing interest on very small satellites >50% failure for satellites <50kg Very limited studies on statistical analysis TUD’s previous work on satellites <500kg

4. 4 Introduction Research Innovations Looks specifically into satellites under 50 kg Specific Focuses more on failures on component level Focus Links statistical analysis with case studies Link

5. 5 Small Satellites Anomalies Database 50 Empirical Data 141 anomalies / 117 satellites

6. 6 Non-parametric Analysis Overall Reliability

7. 7 Non-parametric Analysis Subsystem Contribution

8. 8 Parametric Analysis Weibull Model Probability Density Function Weibull Parameters Shape parameter β: 0.3134 Scale parameter η [days]: 3062

9. 9 Parametric Analysis Subsystem Level SubsystemScale ηShape β ADC859 6070.3013 C&DH29 4470.3925 EPS13 6840.3655 M&S74 895 5260.2643 TT&C28 5200.2513 SubsystemMinor failureMajor failureFatal failure Scale ηShape βScale ηShape βScale ηShape β ADC1 276 7170.4202146 935 0360.24314 539 4870.3849 C&DH119 0400.4025907 0340.3910973 2270.3961 EPS5 814 7560.3851286 0860.398122 8310.3820 M&S16 877 1920.34194.09E+90.2192-- TT&C16 615 4600.26096 394 6830.250083 2990.2841 SeverityFailure characteristics MinorProblem can be fixed from the ground or solved by redundancy. Possible degraded operation but no threat to mission objectives. MajorNon-repairable failure causing partial loss of functionality of the satellite or its subsystems on a permanent basis. FatalTotal loss of functionality of the satellite.

10. 10 Parametric Analysis Component Level ComponentScale ηShape β Actuator3 152 9660.3915 Antenna1.26E+110.1699 Battery25 9850.4667 Deployment2.21E+110.2046 Internal Comm249 6240.4994 Memory39 048 1970.2807 OBC177 9110.6234 Payload9 805 3440.3195 Sensor24 439 8200.3128 Software4 346 081 6710.2529 Solar panel1 881 6450.3657 Thermal50 801 5210.3197 Transceiver342 7230.2748 Unknown29 5470.2863

11. 11 Parametric Analysis Component Level ComponentMinor failureMajor failureFatal failure Scale ηShape βScale ηShape βScale ηShape β Actuator128 130 3810.37546 544 2620.423422 173 0670.4321 Antenna7.99E+130.18232.34E+120.1810-- Battery10 263 7110.405575 4340.814231 2150.4586 Deployment--4.09E+100.2191-- Internal Comm618 4970.439728 5081.6532-- Memory1.06E+100.236536 423 0150.421940 269 0190.3460 OBC28 1751.2850--5 259 3610.4240 Payload--18 818 9070.313311 301 7210.4627 Sensor1 976 2140.42501.13E+150.1806-- Software961 211 4150.2726---- Solar panel306 231 4310.350512 956 7790.39473 100 6350.4027 Thermal16 877 1930.3419---- Transceiver11 089 9590.305360 503 1420.26161 080 7260.3143 Unknown1.51E+90.267419 280 9200.271623 6390.3263

12. 12 Case Study QuakeSat Minor failure CDHS – loss of a multiplexer and interrupted connections at the CPU Major failure EPS – loss of both batteries after 0.5 year Reason Temp caused electrolyte bake out as batteries sealed with non-space packaging Lessons GSE should not always replace batteries during testing COTS components can be used only when fit to space conditions Batteries should be on during thermal vacuum testing Flight-like full testing shall be performed

13. 13 Case Study TUBSAT C Fatal failure EPS – four NiH2 batteries failed after almost 10 years Reason A large number of charge-discharge cycles Lessons An example of a good design and testing of satellite Normally space-rated or specifically designed batteries are better than COTS Full ground testing pays the money back

14. 14 Case Study Falconsat-3 Major failure CDHS – flash memory used to boot CPU was corrupted (solved by using a modified code using only the healthy part of the memory, but need to re-upload flight code every time the CPU was rebooted) ADCS – magnetic sensing and control malfunctioned due to interference (solved with a software modification) Reason CDHS – ground testing found problem but no time to solve ADCS – ground testing for sensor and actuator separately Lessons Testing can easily discover problems, but still need resources to solve problems Integrated testing is necessary to find interference, especially for EMC issues COTS components require additional testing for space qualification (e.g. gravity boom)

15. 15 Battery and transceiver are responsible for most fatal and major failures and deserve more attention Interference is also a main cause of failures COTS components could work well with extensive testing Root cause of failures is either the environment or design flaws Testing is the best way of preventing failures and should not be underestimated Next step is to make the SSAD50 public accessible after information sensitivity check Conclusions

16. 16 For further info, please contact: Dr Jian Guo j.guo@tudelft.nl