Spacecraft attitude control subsystem: Reliability, multi-state analyses, and comparative failure behavior in LEO and GEO

Abstract The attitude control subsystem (ACS) on board a 3-axis stabilized spacecraft is critical for on-orbit operations; its degradation can significantly disrupt mission performance, and its failure is fatal to the spacecraft. An ACS that fails early on-orbit, for example in the first couple of years on a spacecraft with a 15-year design lifetime, results in significant losses to all stakeholders and constitutes a major financial blow to the satellite operator or insurer—losses can amount to a few hundred mission dollars. Understanding and preventing/mitigating ACS failure is an engineering and financial imperative. In this work, we first provide a comprehensive statistical analysis of ACS failure behavior on orbit. The results indicate that the ACS is a major driver of spacecraft unreliability and contributes up to 20% of all spacecraft failures. Nonparametric and parametric results, including mixture Weibull distributions, are provided for ACS reliability, and they are complemented with the subsystem’s multi-state failure analysis. It is shown for example that the ACS is 89% likely to be fully operational after 15 years on-orbit. Results also demonstrate that ACS suffers from both infant mortality and wear-out failures. The findings are important for satellite manufacturers and equipment providers as they indicate opportunities for improved ground testing and burn-in procedures for the ACS, and they highlight the need to consider additional redundancy, even in stand-by form, especially if increased spacecraft design lifetime is sought. Second, we conduct a comparative analysis of failure behavior of the ACS in Low Earth Orbit (LEO) and Geosynchronous orbit (GEO). The results demonstrate a marked difference in failure behavior in LEO and GEO: the ACS degrades and fails more frequently, harder (more severe anomalies), and earlier (more infant anomalies/failures) in LEO than in GEO. We expand on these results and conclude with hypotheses for causal factors of this difference in failure behavior.