ETD

• confidence level
• cryogenic propulsion
• reliability prediction
• uncertainty quantification
• Weibull models

This thesis presents the development and application of a reliability assessment methodology for space propulsion components during the early design and development phases. The main objective of the work is to establish a structured framework capable of predicting component reliability under uncertainty and generating quantitative reliability estimates that can be used as preliminary probabilities to support early design decisions.
The study focuses on cryogenic propulsion systems, with particular emphasis on the Oxygen Main Distribution Valve (OMDV), a fundamental component responsible for enabling or inhibiting oxidizer flow in the MR10 engine.
The core of this work is the development of a Stress–Strength reliability model with uncertain parameters. In this approach, a failure, defined as the inability to perform a required function, is interpreted as the exceedance of a critical threshold. Stress and Strength are modeled as random variables, representing the operational demand and the component capability, respectively, and their realizations define the operational and admissible levels associated with the failure condition. Both Stress and Strength have been described using three-parameter Weibull distributions, while the uncertainties affecting their parameters have been represented through normal distributions. Monte Carlo simulations have been implemented to propagate these uncertainties and to generate probabilistic estimates of the failure probability.
Two different Monte Carlo strategies have been implemented a Classical sampling approach and an innovative Mirror approach, designed to ensure a more balanced representation of the Stress distribution. The convergence behavior of both methods has been analyzed, and their results have been compared. The results show that both Monte Carlo strategies provide comparable estimates of the failure probability, while the Mirror approach yields more conservative results.
Furthermore, since in many practical cases a failure mode is simultaneously influenced by multiple physical quantities, the methodology has been extended from a one-dimensional Stress-Strength model, based on a single physical parameter, to a two-dimensional Stress–Strength formulation that simultaneously accounts for two physical parameters. This two-dimensional approach results useful in evolving scenarios, where new experimental evidence may reveal the influence of additional physical factors not previously considered. The main objective of the two-dimensional method is the evaluation of the joint probability of failure, in order to consider the probability that multiple critical conditions occur simultaneously. Therefore, the two-dimensional analysis reinforces and extends the findings of the one-dimensional model.