Reliability has long been a major consideration in the design of space systems, and in recent years it has become an essential metric in spacecraft design trade-space exploration and optimization. The purpose of this paper is to statistically derive and compare reliability results of Earth-orbiting satellites as a function of orbit type, namely geosynchronous orbits (GEO), low Earth orbits (LEO) and medium Earth orbits (MEO). Using an extensive database of satellite launches and failures/anomalies, life data analyses are conducted over three samples of satellites within each orbit type and successfully launched between 1990 and 2008. Because the dataset is censored, the Kaplan-Meier estimator is used to estimate the reliability functions. Plots of satellite reliability as a function of orbit altitude are provided for each orbit type, as well as confidence bounds on these estimates. Using analytical techniques such as maximum likelihood estimation (MLE), parametric fits are conducted on the previous nonparametric reliability results using single Weibull and mixture distributions. Based on these parametric fits, a comparative reliability analysis is provided identifying similarities and differences in the reliability behaviors of satellites in these three types of orbits. Finally, beyond the statistical analysis, this work concludes with several hypotheses for structural/causal explanations of these trends and difference in on-orbit failure behavior.
展开▼