Probabilistic aerothermal design of gas turbine combustors

Abstract

This thesis presents a probability-based framework for assessing the impact of manufacturing variability on combustor liner durability. Simplified models are used to link combustor liner life, liner temperature variability, and the effects of manufacturing variability. A probabilistic analysis is then applied to the simplified models to estimate the combustor life distribution. The material property and liner temperature variations accounted for approximately 80 percent and 20 percent, respectively, of the combustor life variability. Furthermore, the typical combustor life was found to be approximately 20 percent less than the life estimated using deterministic methods for these combustors, and the probability that a randomly selected combustor will fail earlier than predicted using deterministic methods is approximately 80 percent. Finally, the application of a sensitivity analysis to a surrogate model for the life identified the leading drivers of the minimum combustor life and the typical combustor life as the material property variability and the variability of the near-wall combustor gas temperature, respectively.