Reliability of CubeSats – Statistical Data, Developers’ Beliefs and the Way Forward

In this paper we investigate the data on 178 launched CubeSats and conduct a nonparametric and parametric analysis, where the dead-on-arrival (DOA) cases as well as the subsystem contribution to failure are specifically addressed. Using Maximum Likelihood Estimation, a Single Weibull and a 2-Weibull mixture parametric model are fitted to the non-parametric data. Furthermore, by combining developers’ beliefs on several reliability aspects from a survey conducted in late 2014 with data from past missions, we make a first attempt to correlate space engineering “best guesses” and intuition to actual data. Finally, the probabilistic CubeSat reliability estimation tool is introduced as a method to reduce the infant mortality of CubeSats: CubeSat developers should be able to estimate their required functional testing time on subsystem and system level at an early project stage, while targeting a desired reliability goal on their CubeSat.