Studying reliability: Cryocoolers and power convertors

Abstract

The dynamic machines discussed in this paper are cryocooler and power convertors using various thermodynamic cycles and moving parts. Analyzing dynamic machine reliability has been difficult from the beginning. Classic reliability analyses rely on statistical sampling and comparing failure modes to other similar systems where statistical results are available. These approaches do not apply to this class of dynamic machines, particularly for aerospace applications. The industry has not built enough total machines, let alone a single type of machine, to provide any meaningful statistical sample. This forces us to rely on comparing failure modes of similar systems to that of the machine, which leads to the next problem; today’s aerospace dynamic machine is designed to have no failure modes. What can it be compared to? Any classic reliability study performed on this class of machine makes several critical assumptions that completely dominate the results. Change the assumptions and you get a different answer; the results are dictated by the assumptions not the hardware. There are no easy answers to these problems. This paper attempts to show why classic reliability studies do not apply to aerospace dynamic machines and that as an industry we must work together to show that all cryocoolers have high levels of reliability.