5.1. Scope#

Reliability assessment of mechanical parts in space applications entails several challenges. Firstly, the parts cannot be tested in large numbers and not under all possible operating conditions, thus limiting the use of classical reliability theory. Furthermore, the parts cannot be repaired (except for ground segment equipment), which means that failure of a part may result in mission failure, putting significant pressure on a thorough assessment of parts and systems reliability. These factors, combined with the costs of a spacecraft, provide challenging boundary conditions for engineers in spacecraft design. While classical reliability methods fall short in several ways, an effective answer to these challenges might be found in modern theory of structural reliability as well as in Bayesian probability theory. The designation of structural reliability theory is used in this chapter for methods in which physical processes leading to failures are described with probabilistic models, and reliability is derived from them. In space applications, these methods are also referred to as statistical inference assessment.

Structural reliability theory allows for reliability modelling based on the understanding of the physical processes causing the failures, rather than a large number of tests. The characteristics of mechanical parts, operation conditions, external actions and internal stresses can be considered in detail and in accordance with current state-of-the-art engineering models. Different failure mechanisms and operating conditions can be explicitly modelled, and considered individually or jointly. The use of structural reliability methods for mechanical systems and parts for space structures is detailed in this document. Relevant failure mechanisms are identified and modelling guidelines, including a limit-state function and basic variables modelling is proposed for each. Furthermore, in recognition that these methods should be used on a daily basis, simplified part level methods are developed and presented and interactive models are provided.