This dissertation describes a set of methods used to track, assess and predict the reliability of CubeSats through system level testing and reliability growth modelling. In the last decade, CubeSats have matured from educational tools into accepted scientific and commercial assets, with more than 700 of those standardized satellites launched so far. A recent report by the National Academy of Sciences discussed that CubeSats have shown many characteristics of disruptive innovations, similar to personal computers or cellular phones in the past. However, their current high rates of dead on arrival (DOA) and infant mortality jeopardize this evolution and frequently the underlying cause of this is limited system-level testing done by many developers.
After a definition of the reliability terms and models used, common failures of unmanned past and present spacecraft are described. Although most reliability prediction models use the assumption of random hardware faults, these failure data show that systematic errors, such as failures in design and manufacturing, are the most prevalent source of satellite failure. Also, the increasing utilization of software and the increase in complexity of that software results in more failures in recent missions. The ongoing miniaturization of spacecraft and the professionalization of terrestrial electronics induced the increased use of commercial off-the-shelf (COTS) components for space missions. The qualification of these parts for automotive or industrial purposes make them in many cases also sufficient for space usage, with the exception of vacuum and radiation.
Analyses of current parametric reliability models exposed inconsistencies in the time-dependent behavior of infant mortality and wear-out modelled. New parametric models implemented suggest that the pooled group of satellites of different sizes experiences no distinct wear-out, as their failure rate function has the shape of a right-open bathtub-curve. Splitting the group up into different mass-classes reveals that wear-out is more prevalent in larger satellites, and that the reliability of smaller satellites is dominated by DOA and failures throughout their lifetime. No significant difference can be observed between the reliability of the different sizes of satellites within the observation window. For CubeSats, the on-orbit reliability data were collected from various sources and used to build the so-called CubeSat Failure Database (CFDB). The extraction of the time-dependent failure behavior of this class of satellites proves that DOA and infant mortality are the most prevalent contributors for CubeSat failure.
To prevent future CubeSat missions from experiencing early failure, a reliability assessment method to identify, track, and subsequently solve possible DOA and infant mortality causes was verified on the CubeSat MOVE-II of the Technical University of Munich. The method is based on an adapted reliability growth model and an online, semi-automated Failure Reporting and Corrective Action System (FRACAS). Out of several growth models tested with the failure data of the satellite, the basic exponential and the delayed exponential growth models show the most promising results. Using the growth models, the remaining failures in the system, the space segment and the remaining critical failures, as well as the on-orbit reliability are estimated. Besides this, methods implemented to maximize the number of beta testers interacting with the satellite in a Test Like You Fly (TLYF) configuration and approaches how to shift risk upfront were developed. Finally, for future missions, a reliability prediction method to efficiently trade-off design options in early phases is shown.
«
This dissertation describes a set of methods used to track, assess and predict the reliability of CubeSats through system level testing and reliability growth modelling. In the last decade, CubeSats have matured from educational tools into accepted scientific and commercial assets, with more than 700 of those standardized satellites launched so far. A recent report by the National Academy of Sciences discussed that CubeSats have shown many characteristics of disruptive innovations, similar to pe...
»